Evidence review for preventing contrast-induced acute kidney injury

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Evidence review for preventing contrast-induced acute kidney injury

Acute kidney injury: prevention, detection and management

Evidence review A

NICE Guideline, No. 148

London: National Institute for Health and Care Excellence (NICE); 2019 Dec.

ISBN-13: 978-1-4731-3623-6

Preventing contrast-induced acute kidney injury

Review question

What is the comparative clinical and cost effectiveness of N-acetylcysteine (NAC) and/or fluids in preventing contrast induced acute kidney injury (CI-AKI) in at risk adults?

Introduction

Exposure to iodinated contrast media has been associated with in-hospital AKI. Acute kidney injury following administration of iodinated contrast has previously been referred to as contrast induced nephropathy (CIN). The Kidney Disease: Improving Global Outcomes (KDIGO) international guideline proposes adopting the term contrast induced AKI (CI-AKI) and applying the KDIGO AKI definition. This will provide the opportunity to standardise the terminology used to define AKI and stage its severity. Contrast induced-AKI is uncommon in the general population, with an incidence of 1–2%, and occurs within 72 hours of receiving iodinated contrast media, usually recovering over the following five days. Its incidence increases significantly in patients with risk factors and is associated with prolonged hospital stay, increased mortality and increased health care costs. The risk of CI-AKI has been reported to be as high as 25% in patients with a combination of chronic kidney disease (CKD) and diabetes, cardiac failure, older age and exposure to nephrotoxic drugs.

The NICE guideline on acute kidney injury: prevention, detection and management (NICE guideline CG169) was reviewed in 2017 as part of NICE’s surveillance programme in an exceptional review. The purpose of the exceptional review was to examine any impact on the acute kidney injury guideline following the publication of the AMACING study (Nijssen et al 2017), which compared the effectiveness of no prophylaxis to intravenous volume expansion with 0.9% sodium chloride, in people referred for an elective procedure requiring intravascular-iodinated contrast material who were at high risk of CI-AKI. The results showed non-inferiority of either treatment. This new trial result was seen as potentially sufficient to change the existing recommendations. As a result, the decision was made to update this part of the guideline.

The aim of this review is to assess the clinical and cost effectiveness of NAC and/or fluids in preventing CI-AKI in at risk adults. This review identified randomised controlled trials (RCTs) that fulfilled the conditions specified in Table 1. For full details of the review protocol, see appendix A.

PICO table

Table 1

PICO table for preventing Contrast induced acute kidney injury review.

Methods and process

This evidence review was developed using the methods and process described in Developing NICE guidelines: the manual. Methods specific to this review question are described in the review protocol in appendix A and the methods section in appendix B.

Declarations of interest were recorded according to NICE’s 2018 conflicts of interest policy.

The following methods were specific for this review:

(a)

References that were excluded on sample size (N<80) from the guideline in 2013 for this review question were added back in if they were excluded only on the basis of a sample size N<80 to ensure consistency between the new data set and the original data set.

(b)

Included RCTs reported CI-AKI using different definitions and different time points:

(c)

Table 34 in appendix P shows a list of reported CI-AKI definitions. Some RCTs reported on more than one CI-AKI definition. A ranked list of CI-AKI definitions was developed to prioritise data extraction with the result that only one definition was extracted per trial. The prioritisation was based on committee discussions about which definitions were most clinically useful and the frequency of reporting using each definition in the included RCTs. See appendix B for the prioritisation of CI-AKI definitions. The committee agreed that RCTs reporting different CI-AKI definitions could be analysed together as all these definitions were indicative of CI-AKI.

(d)

RCTs reported CI-AKI events at different time points ranging from 1 to 5 days. The committee agreed that RCTs reporting different time points could be analysed together as long as the longest time point was 5 days or less.

(e)

Included RCTs also reported different regime, duration, and volume (or dosage in the case of NAC) for most of the interventions. Details of included interventions are reported in appendix D clinical evidence tables. The committee agreed that RCTs could be analysed together grouping different regimes, durations, and volumes/dosage within a type of fluid:

sodium chloride 0.9% (IV)
no (intravenous) hydration
sodium bicarbonate (IV)
sodium chloride 0.45% (IV)
sodium citrate (oral)
oral fluids
sodium bicarbonate (oral) + oral fluids
sodium chloride 0.45% (IV) + sodium bicarbonate (IV)
NAC (oral) + sodium chloride 0.9% (IV)
NAC (oral) + sodium bicarbonate (IV)
sodium chloride 0.9% (IV) + sodium bicarbonate (IV)
NAC (IV) + sodium chloride 0.9% (IV)
NAC (oral) + sodium chloride 0.45% (IV)
NAC (IV) + sodium chloride 0.45% (IV)
NAC (oral)
NAC (IV bolus & oral) + sodium chloride 0.9% (IV)
NAC (IV bolus) + sodium chloride 0.9% (IV)

(f)

Some studies reported serum creatinine in mg/l, but the committee highlighted that µmol/l is the preferred unit of meausure in the UK. Therefore, any data on serum creatinine reported in mg/l were converted to the preferred measure µmol/l by multiplying mg/l by 88.4.

(g)

Chen 2008 recruited 2 groups based on their baseline serum creatinine (group 1: serum creatinine <132.6μmol/l; group 2: serum creatinine ≥132.6μmol/l). Group 1 was allocated to sodium chloride 0.45% or no (intravenous) hydration. Group 2 was allocated to oral NAC + sodium chloride 0.45% or oral NAC. Therefore, Chen 2008 was split as 2008a and 2008b in the NMA data.

(h)

Aslanger 2012 stated that sodium chloride 0.9% was given to all participants but the authors did not specify whether this fluid was given pre and/or post contrast. We assumed that the fluid was given at both pre and post contrast because the fluid was given for 12 hours and all people were given contrast within 12 hours of symptom onset, therefore all fluid must have continued after the contrast.

(i)

Adverse events were extracted as number of people with adverse events rather than number of events to enable pooling of RCTs reporting on adverse events. Data was not extracted from RCTs only reporting number of events.

(j)

The NMA models for a dichotomous outcome were based on models from the NICE Decision Support Unit (DSU) technical support document 2 (models 1c and 1d). The models are shown in appendix Q.

(k)

Results were reported as the posterior median and 95% credible interval from the NMA models with the best fit to the data based on the NICE Guideline Updates team criteria for model choice detailed in appendix B.

(l)

The DSU code presents the results of dichotomous outcomes as OR. These were converted to RR by the NICE Guideline Updates Team. Relative effects calculated on a log(odds ratio) scale were re-expressed as relative risks using the absolute probability of CI-AKI (25/250) from the sodium chloride 0.9% + oral NAC arm of Maioli et al. (2008). This study was selected to provide the baseline probability as it is a relatively large, European study that also reports results that are used to estimate the consequences of CI-AKI in the HE model; therefore, using this baseline probability allows for a consistent chain of evidence.

(m)

Where the data for the NMA for CI-AKI (dichotomous outcome) included RCTs with 0 events in both arms, these RCTs were not included as part of the analysis because RCTs with 0 events in both arms do not contribute evidence on the relative treatment effects in pairwise meta-analysis or NMA.

(n)

CI-AKI was reported as defined by study in pairwise analysis (usually 48–72 hours but diagnosed within 7 days of contrast being given to allow for delays in testing). For the NMA, number of diagnoses of CI-AKI within 5 days of the contrast being given was selected as the most appropriate outcome to prioritise because there were sufficient numbers of trials to form a connected network that included the majority of interventions.

(o)

Inconsistency checking of the NMA was carried out (see Appendix R – NMA inconsistency checks).

(p)

Although there were studies at high risk of bias included in the NMA, sensitivity analyses excluding these studies were not carried out because sensitivity analyses for the pairwise data did not alter the interpretation of the effects of the treatments with 2 exceptions (oral NAC + sodium chloride 0.45% compared to sodium chloride 0.45%; oral NAC + sodium chloride 0.9% compared to oral NAC + sodium bicarbonate). These were not considered sufficient to warrant running NMA sensitivity analyses for the CI-AKI outcome. See section on ‘The quality of the evidence’ for more information about the differences in interpretation between the analysis including all studies and the sensitivity analysis removing studies at high risk of bias.

We would like to acknowledge the Technical Support Unit, at University of Bristol, particularly Nicky Welton and Caitlin Daly, for providing advice, models, inconsistency checking and quality assurance for the network meta-analysis included in this review.

Clinical evidence

Included studies

A systematic search was carried out to identify randomised controlled trials (RCTs) and systematic reviews of RCTs, which found 592 references (see appendix C for the literature search strategy). Evidence identified in the original guideline (37 references), excluded references in the original guideline with sample size <80 participants (20 references). References from the NICE surveillance review (29 references), and from systematic reviews (see below) were also reviewed.

In total, 647 references were identified for screening at title and abstract level with 490 excluded at this level. Full texts were ordered to be screened for 157 references (43 systematic reviews and 114 RCTs).

Forty-three systematic reviews were identified in the full text screen. There were 6 network meta-analyses published between 2017 and 2019. None of these network meta-analyses matched the question under consideration here and so a de novo NMA was carried out. The existing network meta-analyses were used as additional sources of references (10 RCTs). In total 75 references (reporting on 70 RCTs) were included based on their relevance to the review protocol (appendix A). The clinical evidence study selection is presented as a PRISMA diagram in appendix D.

See appendix O for a list of references for included studies.

Excluded studies

See appendix M for a list of excluded studies with reasons for exclusion and appendix O for the bibliographic reference.

Table 2

Summary of clinical studies included in the evidence review.

See appendix E for full evidence tables.

Quality assessment of clinical studies included in the evidence review

Network meta-analysis

All analyses are for the outcome CI-AKI as this was the outcome reported by all of the included studies and was the only outcome amenable to NMA. For full GRADE tables see appendix H.

Table 3

Summary GRADE table (outcome: CI-AKI).

Pairwise meta-analysis

All analyses are for the outcome CI-AKI as this was the outcome reported by all of the included studies. For full GRADE tables see appendix H.

Table 4

Summary GRADE table (outcome: CI-AKI).

Table 5

Summary GRADE table; Pre-specified subgroups on pairwise data (outcome: CI-AKI).

Table 6

Summary GRADE table; Other outcomes.

Table 7

Summary GRADE table; Sensitivity analysis on pairwise data excluding studies with a high risk of bias (outcome: CI-AKI).

See appendix H for full GRADE tables.

Economic evidence

A search was conducted to identify economic evaluations relevant to the review question (see Appendix C – Literature search strategies). Search sets covering the original interventions were date limited from January 2013 (when the original search was conducted), while an additional search set covered an expanded version of the fluid therapy terms and was not date limited. The search returned a total of 135 records, 131 of which were excluded on the basis of title and abstract. The remaining 4 studies were fully inspected and none were included in the synthesis. No additional studies were identified during inspection of the full publications and reference lists. The economic evidence study selection is presented as a PRISMA diagram in Appendix I – Economic evidence study selection.

Included studies

No studies were included.

Excluded studies

Details of excluded studies are provided in Appendix M – Excluded studies.

Summary of studies included in the economic evidence review

No economic evaluations relevant to the review question were found.

Economic model

An economic model was developed to answer the review question ‘What is the comparative clinical and cost effectiveness of N-acetylcysteine (NAC) and/or fluids in preventing contrast induced acute kidney injury (CI-AKI) in at risk adults?’. Table 8 presents an economic evidence profile summarising the model. See Appendix L – Health economic analysis for a full model report (including a list of interventions and comparators).

Table 8

Original cost–utility model – economic evidence profile.

The committee’s discussion of the evidence

Interpreting the evidence

The outcomes that matter most

The committee agreed that the key outcome for people at risk of contrast induced acute kidney injury (CI-AKI) was the occurrence of CI-AKI. Committee members highlighted that CI-AKI was normally diagnosed within 24–48 hours after the contrast was given, but had to be diagnosed within 7 days of contrast being given (this was to allow for delays in testing) because later acute kidney injury (AKI) diagnosis was not likely to be related to the use of iodine based contrast media. The committee agreed that chronic kidney disease (CKD) progression, mortality, needing renal replacement therapy, and adverse events (including heart failure) were also important outcomes as these could indicate that an intervention was not working or might be harmful. However, these outcomes were not prioritised because the committee expected that there would be a shortage of evidence, making it harder to use them for decision making.

The quality of the evidence

Overall, the quality of the pairwise evidence varied from high to very low, with the main reasons for downgrading being due to imprecision of the evidence on the relative effectiveness of different fluids at preventing CI-AKI and risk of bias of included studies. In most of the pairwise comparisons, imprecision was considered to be serious (95% confidence interval crossing one end of the defined (minimal clinically important difference) MID interval [0.8, 1.25]) or very serious (95% confidence interval crossing both ends of the defined MID interval [0.8, 1.25]). Risk of bias for some of the included studies was due to lack of detailed report of the randomisation process, lack of report that protocols were pre-registered, and either participants were aware of which intervention were assigned or the assignment of interventions was not well described.

The quality of the evidence was low for the network meta analysis (NMA). The main reasons for downgrading were due to risk of bias of the included studies for the reasons mentioned above.

As a result of most of the evidence being of very low to moderate quality the committee did not feel able to make strong recommendations and instead made ‘consider’ recommendations.

CI-AKI was reported using different definitions (see appendix P for a list of reported CI-AKI definitions). The committee agreed to prioritise definitions for data extraction based on their clinical usefulness (see appendix Q for the prioritisation of CI-AKI definitions). CI-AKI events were reported at different time points ranging from 1 to 5 days. The committee agreed that randomised controlled trials (RCTs) reporting different time points could be analysed together as long as the longest time point was 5 days or less. The committee also agreed that RCTs could be analysed together grouping different regimes, durations, and volumes/dosage within a type of fluid.

The committee highlighted that high-osmolar contrast agents are currently not recommended due to the high risk of adverse reactions. Therefore, it made sure that all included studies used contrast agents that were either low-osmolar or iso-osmolar. There was one study (Solomon 2015) stating that the choice of contrast agent was left to individual participants sites, so the committee were unable to be sure that none of the sites used high osmolar contrast agent.

The committee noted that all included studies reported data on occurrence of CI-AKI, but there was limited evidence for the rest of outcomes (CKD progression, mortality, need for renal replacement therapy, adverse events, hospital stay, readmission for AKI, and health related quality of life). There was also limited evidence on subgroup analyses. The committee noted that none of the subgroup analyses showed evidence of an effect from any of the interventions on the incidence of CI-AKI. Regarding the rest of the outcomes, there was evidence of fewer adverse events with sodium chloride 0.45% compared to oral NAC + sodium chloride 0.45% but the confidence interval crossed the MID; fewer days in hospital with NAC + sodium chloride 0.9% compared to sodium chloride 0.9%; and fewer deaths in-hospital with IV NAC bolus + oral NAC + sodium chloride 0.9% compared to sodium chloride 0.9%. This evidence came from single RCTs and was not compelling compared to the much greater statistical power of the NMA, therefore the committee agreed to make decisions based on the NMA findings.

Sensitivity analyses for the pairwise data did not alter the interpretation of the effects of the treatments with 2 exceptions that were not considered sufficient to warrant running NMA sensitivity analyses for the CI-AKI outcome:

Oral N-acetylcysteine (NAC) + sodium chloride 0.45% compared to sodium chloride 0.45%. Sensitivity analysis showed that oral NAC + sodium chloride 0.45% could not demonstrate a meaningful difference when studies at high risk of bias were removed (previously, a meaningful effect exceeding the MID [0.8, 1.25]).
Oral NAC + sodium chloride 0.9% compared to oral NAC + sodium bicarbonate. Sensitivity analysis showed that oral NAC + sodium chloride 0.9% could not differentiate to oral NAC + sodium bicarbonate when studies at high risk of bias were removed (previously, could not demonstrate a meaningful difference crossing one end of the MID [0.8, 1.25]).

NMA analyses and NMA model inconsistency checks

The NMA model included 17 different interventions. The results of this model showed substantial within-contrast heterogeneity. Therefore, a number of study-level characteristics were explored but no intervention-level differences could explain the heterogeneity. As a next step, other NMA models were built and explored to look for a parsimonious model and to improve clinical interpretability. These models broke down each intervention into its constituent elements:

underlying fluid
- sodium chloride 0.9% (intravenous [IV])
- no (intravenous) hydration
- sodium bicarbonate (IV)
- sodium chloride 0.45% (IV)
- sodium citrate (oral)
- oral fluids
- sodium bicarbonate (oral) + oral fluids
- sodium chloride 0.45% (IV) + sodium bicarbonate (IV)
- sodium chloride 0.9% (IV) + sodium bicarbonate (IV)
whether NAC was given or not (oral or intravenous)
what fluid was given at pre- and post-procedure using iodinated contrast
type of procedure was done
- intervention
- diagnostic
- both
setting
- elective
- emergency
- both

All of the different NMA models that were run had similar heterogeneity and total residual deviance. Therefore the simpler 17 intervention model was reverted to because it made fewer assumptions and had marginally lower heterogeneity compared with the rest of the models.

NMA model inconsistency checks were carried out to assess the consistency assumption in the NMA models used to estimate the comparative clinical and cost effectiveness of NAC and/or fluids in preventing CI-AKI in at risk adults.

Firstly, parts of the network containing the potentially inconsistent studies were identified. The characteristics of the studies identified as being potentially inconsistent were examined in detail to determine if there were any differences between these studies and the other studies in the loop in question that could explain the inconsistency. If substantial differences were identified this might suggest that the potentially inconsistent studies should be excluded from the NMA or placed in a separate/different node in the network. These checks focused on key factors that the committee had previously mentioned during their discussions that could potentially alter the results substantially, such as type of procedure (intervention versus diagnostic) and setting (elective or emergency).

Secondly, the characteristics of the other RCTs within the loops were examined to determine whether any of them could be causing the inconsistency instead. In both cases, no differences in study characteristics were identified that could account for the inconsistency and therefore there were no reasons to exclude any of the individual studies.

Thirdly, the NMA model was re-run without the potentially inconsistent study (Ueda 2011) to investigate the effect this study had on the NMA results. This analysis showed minor differences in results compared to the original NMA which included Ueda 2011.

Finally, the NMA model including 17 different interventions and the inconsistency checks were used to interpret the results related to the occurrence of CI-AKI. These results were used by the committee in conjunction with the outcomes of the health economic model when it discussed the benefits and harms of the different interventions in preventing CI-AKI in at risk adults (see next section which includes the discussion of the committee).

This information allowed the committee to discuss the relative effectiveness of all of the combinations of NAC and fluids compared to each other and therefore it was able to be clearer about their recommendations even though the quality of the evidence ws not sufficient to make strong recommendations.

Benefits and harms

Occurrence of CI-AKI was similar across interventions (either oral or intravenous fluids) in the NMA and there was limited evidence for other outcomes and subgroup analyses from the pairwise data.

The committee agreed that outpatients are generally at lower risk of CI-AKI compared with inpatients with particular risk factors (acutely ill, estimated glomerular filtration rate [eGFR] < 30 ml/min/1.73 m², renal transplant, increased volume of contrast agent, intra-arterial administration of contrast agent) so it made separate recommendations for each group of patients. Most of the risk factors given above were taken from recommendation 1.1.6 of the current guideline apart from the level of eGFR which was based on the committee’s clinical knowledge and experience. The committee agreed that a level of eGFR < 30 ml/min/1.73 m² was appropriate. The committee noted that the evidence was unclear about the effectiveness of intravenous/oral fluids stratified by different eGFR subgroups on the incidence of CI-AKI. It made a research recommendation to investigate whether the risk of CI-AKI can be stratified by eGFR.. It noted that it might not be possible to do this research on people with very low eGFRs because they may be too high risk to be included in research studies. However it agreed that better evidence of risk stratification for CI-AKI in people with higher eGFRs would still improve clinical practice and patient safety. The committee also agreed that, based on their experience and expertise, the risk for intra-arterial administration depends on the site of the injection, and is particularly high with first-pass renal exposure, because the contrast medium passes into the kidneys relatively undiluted. The committee agreed to be more specific about what constituted a ‘large volume’ of contrast agent, and while it agreed that there is no simple definition for this, a useful heuristic is to consider doses higher than a standard diagnostic dose as a ‘large volume’

The committee noted the evidence showed oral fluids were not worse than intravenous fluids for preventing CI-AKI and, on that basis, it agreed that it did not seem necessary to bring outpatients into hospital to give them intravenous fluids prior to receiving iodine based contrast media. Therefore, the committee recommended the use of oral hydration in these patients. The committee highlighted the importance of adequate hydration in people having intravenous iodine based contrast media. The committee discussed the different regimens used in the included studies, but decided that it should not recommend a specific regimen as this would need to be adapted to people’s situations (for example, people with heart failure, age [some frail older people might be less likely to be able to follow an oral fluids regimen if the volume of fluid is high], other conditions [people might have gastric problems preventing them to drink the full amount of oral fluids]) and preferences (some people might prefer to drink tea or coffee as well as water). The different oral regimens seen in the evidence presented to the committee were:

patients were encouraged to drink as much spring or tap water as possible 12 hours before and 12 hours after the procedure (Akyuz 2014)
500 mL of water 4 h prior to contrast exposure stopping 2 h prior to procedure and 600 mL of water post procedure (Cho 2010)
oral mineral water or boiled water (1ml/kg/h) 6 to 12 hours before the procedure and 12 hours after the procedure (Wrobel 2010)

Similarly, the committee agreed that many inpatients could be encouraged to hydrate orally before and after being given a contrast agent, however, the committee agreed that inpatients at particularly high risk should receive intravenous fluids for volume expansion when having a contrast agent. It noted the importance of maintaining the correct fluid balance in these patients because fluids (oral or intravenous) might be harmful for some people leading to fluid overload and cardiovascular events. The evidence from the NMA showed that sodium chloride 0.9% and sodium bicarbonate appear to be equivalent for preventing CI-AKI. Therefore the committee recommended the use of intravenous volume expansion for these patients and kept the recommended interventions from the previous guideline: isotonic sodium bicarbonate or sodium chloride 0.9%.

The committee also highlighted the importance of adequate hydration in all inpatients having iodine based contrast media by having their hydration level assessed, and that safety was particulary important with inpatients at high risk of CI-AKI. The committee discussed some of the contraindications for volume expansion (for example, hypervolemic hyponatremia or active decompensated heart failure [these were some of the exclusion criteria listed in Akyuz 2014]) but it did not make a recommendation about this because it agreed that clinical judgment was the key factor in these cases.

The committee also noted that the NMA did not show evidence for the use of NAC either oral or intravenous and that NAC is not routinely used in clinical practice. Therefore, this intervention was not recommended, however due to the poor quality of the evidence, the committee did not feel able to make a ‘do not offer’ recommendation for NAC and recommended it be included in future research. The rank probability histogram for sodium bicarbonate (oral) + oral fluids showed that this treatment had a probability of around 65% of being the best treatment but histograms were associated with a high degree of uncertainty.

The committee highlighted that it was crucial that imaging should not be delayed purely for volume expansion and it asked for this to be made clearer in editorial refresh of recommendation 1.1.6 which includes a list of risk factors for CI-AKI. The last sentence in that recommendation (‘Ensure that risk assessment does not delay emergency imaging’) has been brought to the start of the recommendation (see full guideline recommendation 1.1.6).

The committee clarified that renal transplant patients were excluded from the evidence of this review but they are mentioned in the updated recommendation as a group at high risk of CI-AKI.

The committee agreed that it was important to discuss with a nephrology team before offering iodine based contrast media to adults on renal replacement therapy including people with kidney transplant. Committee members did not consider it necessary to routinely have this discussion about people with other contraindications to intravenous fluids because it agreed that this decision was better made by individual clinicians. Therefore, the committee agreed to remove other contraindications to intravenous fluids from the recommendation.

Cost effectiveness and resource use

The committee discussed the economic evidence relating to the use of N-acetylcysteine (NAC) and/or fluids in preventing contrast induced acute kidney injury (CI-AKI) in at risk adults. As no published economic evaluations that were relevant to the review question had been found, the committee’s discussion focused on the economic model that was developed for this guideline to be directly applicable to the decision problem. The results of the model were presented to the committee, including probabilistic and deterministic sensitivity analyses and scenario analyses according to elective and emergency presentations (varying baseline risk of CI-AKI and costing assumptions). The committee understood that the model outputs directly reflect the results of the NMA, and the limitations of the NMA discussed above should be kept in mind when considering the model results.

Cost-effectiveness results indicated that sodium bicarbonate (oral) with oral fluids dominates all other interventions. This was the case across the base case, sensitivity and subgroup analyses. However, the committee was cautious of this result given the prior discussions of the NMA evidence; the credible interval surrounding the point estimate for sodium bicarbonate (oral) with oral fluids was very wide, and there was only a single trial arm (comprising 21 participants) contributing to the evidence base. The committee ruled out recommending this intervention as it could not draw any conclusions about its effectiveness due to the high degree of uncertainty in the evidence.

Discussions then moved on to the other interventions. The only regimen that, at its point estimate, is associated with fewer episodes of AKI than oral fluids alone is sodium chloride 0.9% with sodium bicarbonate (IV). Therefore, the committee was interested to see how cost-effectiveness results for this strategy compared with oral fluids. In focusing on this comparison, the committee was not attempting to assess the cost effectiveness of the single regimen of sodium chloride 0.9% with sodium bicarbonate (IV); rather, committee members were interested to explore how the best-performing of all intravenous regimens compared with oral fluids alone, as this gave an indication of the best value that could be gained from an intravenous hydration strategy. Similarly, the committee also found it helpful to review the results of grouped probabilistic sensitivity analyses. It noted that almost half the strategies simulated (8 of 17) are consistent with the existing recommendation – that is, they include intravenous sodium chloride 0.9%, sodium bicarbonate or both. The committee understood that, in the presence of substantial uncertainty, this had the effect of dividing the probability mass thinly between several options that cannot be differentiated. As a result, while any 1 of the strategies has a low probability of representing the optimal balance of costs and benefits, there is a much higher chance that 1 or other of them provides best value. Therefore, the committee found it helpful to review the outputs of probabilistic analyses that broke results from 17 strategies down into a simple 3-way split: (i) oral fluids alone, (ii) intravenous regimens with sodium chloride 0.9% and/or sodium bicarbonate (as currently recommended), (iii) other options (including oral NAC alone, no hydration regimen and IV sodium chloride 0.45%).

In the base case, sodium chloride 0.9% with sodium bicarbonate (IV) has an ICER of £510,922 per QALY compared with oral fluids. All other interventions are dominated. The committee agreed that this reinforced the results of the NMA – that, for the average person undergoing a contrast-enhanced scan, there is no evidence that, when compared with careful oral hydration, an intravenous regimen provides meaningful benefit.

The committee then reviewed scenario analyses that sought to establish whether the balance of benefits, harms and costs was different in different groups of people – in particular, those undergoing elective scans and those being treated in emergency settings. The major distinction between these 2 scenarios is the baseline risk of CI-AKI. To reflect this, CI-AKI rates from the sodium chloride 0.9% arms of trials from emergency and elective settings were synthesised separately to obtain setting-specific baseline event-rates. In addition, costing assumptions varied between the 2 scenarios: in the elective setting, the analysis assumed that people receiving preoperative intravenous infusions would have to be admitted for up to a day before the procedure; in the emergency setting, it was assumed that people would already be inpatients, so the administration of intravenous fluids would not, by itself, be associated with additional time in hospital.

The committee saw that, for elective patients, intravenous treatment appeared to be even worse value for money than in the base case: the ICER for sodium chloride 0.9% with sodium bicarbonate (IV) versus oral fluids was £655,323 per QALY. In contrast, the higher risk of AKI in emergency settings, coupled with the lower marginal cost of intravenous regimens when extra hospital stay is not relevant, led to results suggesting there may be a cost-effective role for intravenous hydration. In this scenario, the ICER for sodium chloride 0.9% with sodium bicarbonate (IV) versus oral fluids reduced to £16,112 per QALY. However, the committee understood that this result was subject to substantial uncertainty. It saw deterministic sensitivity analysis showing that outputs could be meaningfully affected by plausible variations to a range of parameters – including but not limited to the relative effects of 1 or both of sodium chloride 0.9% with sodium bicarbonate (IV) and oral fluids. Probabilistic results suggested that there is about a 60% chance that 1 or other of the simulated regimens containing intravenous sodium chloride 0.9% and/or intravenous sodium bicarbonate provides best value in the emergency setting.

The committee interpreted these results as showing that, when baseline risks are high, intravenous volume expansion may slightly attenuate the risk of CI-AKI and this may be cost effective, so long as it is not necessary to admit the person for the sole purpose of preparing for their procedure in this way. Committee members agreed that these results are consistent with their experience, but also noted that the analyses are conservative in estimating the benefits of oral fluid regimens: while the costs of admission for intravenous volume expansion are accounted for in the model, it cannot capture other disadvantages of unnecessary hospitalisation, including inconvenience for the patient, and increases in common risks (for example, falls and hospital-acquired infections).

For these reasons, the committee agreed that careful attention to oral hydration should be adequate for all outpatient procedures and many inpatient ones, with intravenous volume expansion reserved for cases at particularly high risk of CI-AKI.

When deciding on which IV intervention to recommend in the high risk population, the committee discussed that although sodium chloride 0.9% with sodium bicarbonate (IV) was the most cost-effective, there were concerns about its practical implementation. They advised that pre-mixed sodium chloride and sodium bicarbonate solutions are not available in the UK and would need to be made up by hospital staff, with additional resource and cost implications. The grouped probabilistic results indicate that regimens containing intravenous sodium chloride 0.9% and/or intravenous sodium bicarbonate provide best value in the emergency setting, and individual packs of sodium chloride 0.9% or sodium bicarbonate (IV) can be easily obtained without the same practical issues.

Other factors the committee took into account

The committee agreed that the evidence was probably sufficient to make negative (‘do not offer’) recommendations with respect to NAC and sodium chloride 0.45%. It discussed whether there would be value in doing so, but agreed that these interventions are seldom used in NHS practice, so there is little to be gained by advising practitioners against choosing them. Instead, the committee focused on its positive recommendations about what should be done.

The committee did not consider any evidence relating to the use of ACE inhibitors and ARBs in people having iodine based contrast media and therefore it was unable to update this recommendation..

Appendices

Appendix A. Review protocols

Review protocol for preventing contrast induced acute kidney injury in at risk adults

Download PDF (268K)

Appendix B. Methods

Prioritisation of CI-AKI definitions for pairwise and network meta-analyses

List of CI-AKI definitions to prioritise for pairwise and network meta-analyses

Evidence synthesis and meta-analyses

Where possible, meta-analyses were conducted to combine the results of quantitative studies for each outcome. For continuous outcomes analysed as mean differences, where change from baseline data were reported in the trials and were accompanied by a measure of spread (for example standard deviation), these were extracted and used in the meta-analysis. Where measures of spread for change from baseline values were not reported, the corresponding values at study end were used and were combined with change from baseline values to produce summary estimates of effect. These studies were assessed to ensure that baseline values were balanced across the treatment groups; if there were significant differences at baseline these studies were not included in any meta-analysis and were reported separately.

Evidence of effectiveness of interventions

Quality assessment

Individual RCTs were quality assessed using the Cochrane Risk of Bias (RoB) Tool version 2.0. individual study was classified into one of the following three groups:

Low risk of bias – The true effect size for the study is likely to be close to the estimated effect size.
Some concerns – There is a possibility the true effect size for the study is substantially different to the estimated effect size.
High risk of bias – It is likely the true effect size for the study is substantially different to the estimated effect size.

Each individual study was also classified into one of three groups for directness, based on if there were concerns about the population, intervention, comparator and/or outcomes in the study and how directly these variables could address the specified review question. Studies were rated as follows:

Direct – No important deviations from the protocol in population, intervention, comparator and/or outcomes.
Partially indirect – Important deviations from the protocol in one of the population, intervention, comparator and/or outcomes.
Indirect – Important deviations from the protocol in at least two of the following areas: population, intervention, comparator and/or outcomes.

Methods for combining intervention evidence (pairwise analysis)

Meta-analyses of interventional data were conducted with reference to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins et al. 2011).

Where different studies presented continuous data measuring the same outcome but using different numerical scales (e.g. a 0–10 and a 0–100 visual analogue scale), these outcomes were all converted to the same scale before meta-analysis was conducted on the mean differences. Where outcomes measured the same underlying construct but used different instruments/metrics, data were analysed using standardised mean differences (Hedges’ g).

A pooled relative risk was calculated for dichotomous outcomes (using the Mantel–Haenszel method) reporting numbers of people having an event. Both relative and absolute risks were presented, with absolute risks calculated by applying the relative risk to the pooled risk in the comparator arm of the meta-analysis (all pooled trials).

Fixed- and random-effects models (der Simonian and Laird) were fitted for all syntheses, with the presented analysis dependent on the degree of heterogeneity in the assembled evidence. Fixed-effects models were the preferred choice to report, but in situations where the assumption of a shared mean for fixed-effects model were clearly not met, even after appropriate pre-specified subgroup analyses were conducted, random-effects results are presented. Fixed-effects models were deemed to be inappropriate if one or both of the following conditions was met:

Significant between study heterogeneity in methodology, population, intervention or comparator was identified by the reviewer in advance of data analysis. This decision was made and recorded before any data analysis was undertaken.
The presence of significant statistical heterogeneity in the meta-analysis, defined as I²≥50%.

In any meta-analyses where some (but not all) of the data came from studies at high risk of bias, a sensitivity analysis was conducted, excluding those studies from the analysis. Results from both the full and restricted meta-analyses are reported. Similarly, in any meta-analyses where some (but not all) of the data came from indirect studies, a sensitivity analysis was conducted, excluding those studies from the analysis.

Meta-analyses were performed in Cochrane Review Manager V5.3.

Minimal clinically important differences (MIDs)

The Core Outcome Measures in Effectiveness Trials (COMET) database was searched to identify published minimal clinically important difference thresholds relevant to this guideline. Identified MIDs were assessed to ensure they had been developed and validated in a methodologically rigorous way, and were applicable to the populations, interventions and outcomes specified in this guideline. In addition, the Guideline Committee were asked to prospectively specify any outcomes where they felt a consensus MID could be defined from their experience. In particular, any questions looking to evaluate non-inferiority (that one treatment is not meaningfully worse than another) required an MID to be defined to act as a non-inferiority margin. However, no consensus MIDs were defined and no published MIDs were found.

For standardised mean differences where no other MID was available, an MID of 0.2 was used, corresponding to the threshold for a small effect size initially suggested by Cohen et al. (1988). For mean differences where no other MID was available, an MID of +/− 0.5 standard deviations from the mean value was used (Norman 2003). For relative risks where no other MID was available, a default MID interval for dichotomous outcomes of 0.8 to 1.25 was used.

MIDs were used both for assessing imprecision in GRADE and also for assessing clinical importance of treatment effects.

When decisions were made in situations where MIDs were not available, the ‘Evidence to Recommendations’ section of that review should make explicit the committee’s view of the expected clinical importance and relevance of the findings. In particular, this includes consideration of whether the whole effect of a treatment (which may be felt across multiple independent outcome domains) would be likely to be clinically meaningful, rather than simply whether each individual sub outcome might be meaningful in isolation.

GRADE for pairwise meta-analyses of interventional evidence

GRADE was used to assess the quality of evidence for the selected outcomes as specified in ‘Developing NICE guidelines: the manual (2014)’. Data from RCTs was initially rated as high quality and the quality of the evidence for each outcome was downgraded or not from this initial point, based on the criteria given in Table 9

Table 9. Rationale for downgrading quality of evidence for intervention studies

The quality of evidence for each outcome was upgraded if any of the following three conditions were met:

Data from non-randomised studies showing an effect size sufficiently large that it cannot be explained by confounding alone.
Data showing a dose-response gradient.
Data where all plausible residual confounding is likely to increase our confidence in the effect estimate.

Publication bias

Publication bias was assessed where 10 or more studies were included as part of a single meta-analysis and a funnel plot was produced to graphically assess the potential for publication bias^a.

Methods for combining direct and indirect evidence (network meta-analysis) for interventions

Conventional ‘pairwise’ meta-analysis involves the statistical combination of direct evidence about pairs of interventions that originate from two or more separate studies (for example, where there are two or more studies comparing A vs B).

In situations where there are more than two interventions, pairwise meta-analysis of the direct evidence alone is of limited use. This is because multiple pairwise comparisons need to be performed to analyse each pair of interventions in the evidence, and these results can be difficult to interpret. Furthermore, direct evidence about interventions of interest may not be available. For example studies may compare A vs B and B vs C, but there may be no direct evidence comparing A vs C. Network meta-analysis overcomes these problems by combining all evidence into a single, internally coherent model, synthesising data from direct and indirect comparisons, and providing estimates of relative effectiveness for all comparators and the ranking of different interventions. Network meta-analyses were undertaken in all situations where the following three criteria were met:

At least three treatment alternatives.
A sufficiently connected network to enable valid estimates to be made.
The aim of the review was to produce recommendations on the most effective option, rather than simply an unordered list of treatment alternatives.

Synthesis

Hierarchical Bayesian Network Meta-Analysis (NMA) was performed using WinBUGS version 1.4.3. The models used reflected the recommendations of the NICE Decision Support Unit’s Technical Support Documents (TSDs) on evidence synthesis, particularly TSD 2 (‘A generalised linear modelling framework for pairwise and network meta-analysis of randomised controlled trials’; see http://www.nicedsu.org.uk) with additional models provided by the TSU (see appendix O for NMA models).

Results were reported summarising 50,000 samples from the posterior distribution of each model, having first run and discarded 50,000 ‘burn-in’ iterations. Three separate chains with different initial values were used.

Non-informative prior distributions were used in all models. Unless otherwise specified, trial-specific baselines and treatment effects were assigned Normal(0,10000) priors, and the between-trial standard deviations used in random-effects models were given Uniform(0,5) priors. These are consistent with the recommendations in TSD 2 for dichotomous outcomes.

Fixed- and random-effects models were explored for each outcome, with the final choice of model based on deviance information criterion (DIC): if DIC was at least 3 points lower for the random-effects model, it was preferred; otherwise, the fixed effects model was considered to provide an equivalent fit to the data in a more parsimonious analysis, and was preferred.

In any meta-analyses where some (but not all) of the data came from studies at high risk of bias, a sensitivity analysis was conducted, excluding those studies from the analysis. Results from both the full and restricted meta-analyses are reported. Similarly, in any meta-analyses where some (but not all) of the data came from studies that were partially or indirectly applicable compared to the protocol, a sensitivity analysis was conducted, excluding those studies from the analysis. Where sufficient studies were available, meta-regression was undertaken to explore the effect of study level covariates.

Choice of outcomes for network meta-analysis

Number of diagnoses of CI-AKI within 5 days of the an iodine based contrast media being given intravenously or intra-arterially was selected as the most appropriate outcome to prioritise because there were sufficient numbers of trials to form a connected network that included the majority of interventions.

Modified GRADE for network meta-analyses

A modified version of the standard GRADE approach for pairwise interventions was used to assess the quality of evidence across the network meta-analyses undertaken (Table 10). While most criteria for pairwise meta-analyses still apply, it is important to adapt some of the criteria to take into consideration additional factors, such as how each ‘link’ or pairwise comparison within the network applies to the others. As a result, the following was used when modifying the GRADE framework to a network meta-analysis. It is designed to provide a single overall quality rating for an NMA, which can then be combined with pairwise quality ratings for individual comparisons (if appropriate), to judge the overall strength of evidence for each comparison.

Table 10. Rationale for downgrading quality of evidence for intervention studies

Appendix C. Literature search strategies

RQ: What is the comparative clinical and cost effectiveness of N-acetylcysteine (NAC) and/or fluids in preventing contrast induced acute kidney injury (CI-AKI) in at risk adults?

Sources searched to identify the clinical evidence

Databases	Date searched	Version/files	No. retrieved
Cochrane Central Register of Controlled Trials (CENTRAL)	1/04/2019	Cochrane Central Register of Controlled Trials Issue 4 of 12, April 2019	446
Cochrane Database of Systematic Reviews (CDSR)	1/04/2019	Cochrane Database of Systematic Reviews Issue 4 of 12, April 2019	11
Database of Abstracts of Reviews of Effect (DARE)	1/04/2019	CRD	16
Embase (Ovid)	8/04/2019	Embase 1974 to 2019 Week 14	396
MEDLINE (Ovid)	8/04/2019	Ovid MEDLINE(R) 1946 to April 05, 2019	277
MEDLINE In-Process (Ovid)	1/04/2019	Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations 1946 to March 29, 2019	37
MEDLINE Epub Ahead of Print	1/04/2019	Ovid MEDLINE(R) Epub Ahead of Print March 29, 2019	4
Health Technology Assessment (HTA Database)	1/04/2019	CRD	0
Total after de-duplication			592

Search strategies

Database: Medline
Strategy used:
1 exp Acute Kidney Injury/ (42761)
2 (contrast-induced* or radiocontrast-induced* or ci).tw. (432886)
3 1 and 2 (2583)
4 Contrast Media/ae (9289)
5 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*).tw. (8829)
6 (contrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (529)
7 (radiocontrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (40)
8 (contrast* adj4 (nephropath* or nephrotoxi*)).tw. (2580)
9 (radiocontrast* adj4 (nephropath* or nephrotoxi*)).tw. (297)
10 or/3–9 (19554)
11 Acetylcysteine/ (12581)
12 Sodium Chloride/ (57460)
13 Bicarbonates/ (21465)
14 Saline Solution, Hypertonic/ (5470)
15 Saline*.tw. (152324)
16 (Nacet* or acet* or n-acet* or parvolex or mucomyst).tw. (514653)
17 (sodium adj2 (chloride* or bicarbonat*)).tw. (22711)
18 or/11–17 (737710)
19 exp Fluid Therapy/ (19118)
20 Electrolytes/ (24768)
21 (balance* adj2 electrolyte* adj2 solution*).tw. (201)
22 fluid*.tw. (390025)
23 oral* rehydrat*.tw. (2680)
24 (Hartmann* or PlasmaLyte).tw. (3287)
25 or/19–24 (425775)
26 (MEDLINE or pubmed).tw. (137811)
27 systematic review.tw. (97376)
28 systematic review.pt. (103766)
29 meta-analysis.pt. (99181)
30 intervention*.ti. (110298)
31 or/26–30 (327427)
32 randomized controlled trial.pt. (479116)
33 randomi?ed.mp. (737113)
34 placebo.mp. (184329)
35 or/32–34 (786792)
36 31 or 35 (1019975)
37 and/10,18,36 (517)
38 (201301* or 201302* or 201303* or 201304* or 201305* or 201306* or 201307* or 201308* or 201309* or 201310* or 201311* or 201312* or 2014* or 2015* or 2016* or 2017* or 2018* or 2019*).ed. (5209038)
39 37 and 38 (182)
40 and/10,25,36 (178)
41 39 or 40 (306)
42 limit 41 to english language (285)
43 animals/ not humans/ (4533573)
44 42 not 43 (277)

Database: MiP/Epubs
Strategy used:
1 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*).tw. (139)
2 (contrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (16)
3 (radiocontrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (1)
4 (contrast* adj4 (nephropath* or nephrotoxi*)).tw. (43)
5 (radiocontrast* adj4 (nephropath* or nephrotoxi*)).tw. (3)
6 or/1–5 (164)
7 Saline*.tw. (2018)
8 (Nacet* or acet* or n-acet* or parvolex or mucomyst).tw. (7274)
9 (sodium adj2 (chloride* or bicarbonat*)).tw. (303)
10 or/7–9 (9464)
11 (balance* adj2 electrolyte* adj2 solution*).tw. (1)
12 fluid*.tw. (7613)
13 oral* rehydrat*.tw. (17)
14 (Hartmann* or PlasmaLyte).tw. (54)
15 or/11–14 (7676)
16 (MEDLINE or pubmed).tw. (6225)
17 systematic review.tw. (5697)
18 systematic review.pt. (17)
19 meta-analysis.pt. (5)
20 intervention*.ti. (3732)
21 or/16–20 (12240)
22 randomized controlled trial.pt. (1)
23 randomi?ed.mp. (12537)
24 placebo.mp. (3006)
25 or/22–24 (13615)
26 21 or 25 (22956)
27 and/6,10,26 (4)
28 (201301* or 201302* or 201303* or 201304* or 201305* or 201306* or 201307* or 201308* or 201309* or 201310* or 201311* or 201312* or 2014* or 2015* or 2016* or 2017* or 2018* or 2019*).dt. (285027)
29 27 and 28 (4)
30 and/6,15,26 (1)
31 29 or 30 (4)
32 limit 31 to english language (4)

Database: Embase
Strategy used:
1 exp acute kidney failure/ (71966)
2 exp acute kidney tubule necrosis/ (4531)
3 1 or 2 (75020)
4 (contrast-induced* or radiocontrast-induced* or ci).tw. (746179)
5 3 and 4 (6655)
6 contrast medium/ae (5571)
7 contrast induced nephropathy/ (4314)
8 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*).tw. (14398)
9 (contrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (1045)
10 (radiocontrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (60)
11 (radiocontrast* adj4 (nephropath* or nephrotoxi*)).tw. (357)
12 or/5–11 (27080)
13 acetylcysteine/ (33716)
14 sodium chloride/ (171848)
15 bicarbonate/ (45173)
16 hypertonic solution/ (3912)
17 Saline*.tw. (222699)
18 (Nacet* or acet* or n-acet* or parvolex or mucomyst).tw. (693221)
19 (sodium adj2 (chloride* or bicarbonat*)).tw. (30535)
20 or/13–19 (1044216)
21 exp infusion fluid/ (29634)
22 oral rehydration therapy/ (2559)
23 electrolyte/ (36221)
24 acetic acid plus gluconate sodium plus magnesium chloride plus potassium chloride plus sodium chloride/ (312)
25 (balance* adj2 electrolyte* adj2 solution*).tw. (254)
26 fluid*.tw. (544967)
27 oral* rehydrat*.tw. (3031)
28 (Hartmann* or PlasmaLyte).tw. (5022)
29 or/21–28 (605613)
30 (MEDLINE or pubmed).tw. (215134)
31 exp systematic review/ or systematic review.tw. (239844)
32 meta-analysis/ (159213)
33 intervention*.ti. (175512)
34 or/30–33 (559086)
35 random:.tw. (1390441)
36 placebo:.mp. (428367)
37 double-blind:.tw. (195979)
38 or/35–37 (1636073)
39 34 or 38 (2019847)
40 and/12,20,39 (1017)
41 (201301* or 201302* or 201303* or 201304* or 201305* or 201306* or 201307* or 201308* or 201309* or 201310* or 201311* or 201312* or 2014* or 2015* or 2016* or 2017* or 2018* or 2019*).dc. (9825657)
42 40 and 41 (426)
43 and/12,29,39 (234)
44 42 or 43 (600)
45 limit 44 to english language (582)
46 limit 45 to (conference abstract or conference paper) (172)
47 45 not 46 (410)
48 nonhuman/ not (human/ and nonhuman/) (4339559)
49 47 not 48 (396)

Database: Cochrane Library (CDSR and CENTRAL)
Strategy used:
#1 [mh “Acute Kidney Injury”]
#2 (contrast-induced* or radiocontrast-induced* or ci):ti,ab
#3 #1 and #2
#4 MeSH descriptor: [Contrast Media] this term only and with qualifier(s): [adverse effects - AE]
#5 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*):ti,ab
#6 (contrast-induced near/4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)):ti,ab
#7 (radiocontrast-induced near/4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)):ti,ab
#8 (contrast* near/4 (nephropath* or nephrotoxi*)):ti,ab
#9 (radiocontrast* near/4 (nephropath* or nephrotoxi*)):ti,ab
#10 {OR #3-#9}
#11 [mh ^Acetylcysteine]
#12 [mh ^“Sodium Chloride”]
#13 [mh ^Bicarbonates]
#14 [mh ^“Saline Solution, Hypertonic”]
#15 Saline*:ti,ab
#16 (Nacet* or acet* or n-acet* or parvolex or mucomyst):ti,ab
#17 (sodium near/2 (chloride* or bicarbonat*)):ti,ab
#18 {OR #11-#17}
#19 [mh “Fluid Therapy”]
#20 [mh ^Electrolytes]
#21 (balance* near/2 electrolyte* near/2 solution*):ti,ab
#22 fluid*:ti,ab
#23 oral* rehydrat*:ti,ab
#24 (Hartmann* or PlasmaLyte):ti,ab
#25 {OR #19-#24}
#26 #10 and #18 with Cochrane Library publication date Between Jan 2013 and Apr 2019
#27 #10 and #25
#28 #26 or #27
#29 “conference”:pt
#30 #28 not #29

Database: DARE/HTA
Strategy used:
1 MeSH DESCRIPTOR acute kidney injury EXPLODE ALL TREES
2 ((contrast-induced* or radiocontrast-induced* or ci))
3 #1 AND #2
4 MeSH DESCRIPTOR contrast media WITH QUALIFIER AE
5 ((ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*))
6 (contrast-induced) AND ((aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*))
7 (radiocontrast-induced) AND ((aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*))
8 (contrast) AND ((nephropath* or nephrotoxi*))
9 (radiocontrast) AND ((nephropath* or nephrotoxi*))
10 #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9
11 MeSH DESCRIPTOR Acetylcysteine
12 MeSH DESCRIPTOR Sodium Chloride
13 MeSH DESCRIPTOR Bicarbonates
14 MeSH DESCRIPTOR Saline Solution, Hypertonic
15 (Saline*)
16 ((Nacet* or acet* or n-acet* or parvolex or mucomyst))
17 (Sodium) AND ((chloride* or bicarbonat*))
18 #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17
19 MeSH DESCRIPTOR Fluid therapy EXPLODE ALL TREES
20 MeSH DESCRIPTOR Electrolytes
21 (balance) AND (electrolyte) AND (solution*)
22 (fluid*)
23 (oral* rehydrat*)
24 ((Hartmann* or PlasmaLyte))
25 #19 OR #20 OR #21 OR #22 OR #23 OR #24
26 #10 AND #18
27 (#26) IN DARE, HTA FROM 2013 TO 2019
28 #10 AND #25
29 (#28) IN DARE, HTA
30 #27 OR #29

Sources searched to identify economic evaluations

Economics	Date searched
MEDLINE (Ovid)	9th April 2019
MEDLINE in Process (Ovid)	9th April 2019
MEDLINE Epub Ahead of Print	9th April 2019
Embase (Ovid)	9th April 2019
EconLit (Ovid)	9th April 2019
NHS Economic Evaluation Database (NHS EED) (legacy database)	9th April 2019

Database: Medline
Strategy used:
1 exp Acute Kidney Injury/ (42766)
2 (contrast-induced* or radiocontrast-induced* or ci).tw. (433052)
3 1 and 2 (2583)
4 Contrast Media/ae (9291)
5 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*).tw. (8830)
6 (contrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (529)
7 (radiocontrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (40)
8 (contrast* adj4 (nephropath* or nephrotoxi*)).tw. (2581)
9 (radiocontrast* adj4 (nephropath* or nephrotoxi*)).tw. (297)
10 or/3–9 (19556)
11 Acetylcysteine/ (12582)
12 Sodium Chloride/ (57464)
13 Bicarbonates/ (21466)
14 Saline Solution, Hypertonic/ (5470)
15 Saline*.tw. (152339)
16 (Nacet* or acet* or n-acet* or parvolex or mucomyst).tw. (514734)
17 (sodium adj2 (chloride* or bicarbonat*)).tw. (22715)
18 or/11–17 (737812)
19 exp Fluid Therapy/ (19121)
20 Electrolytes/ (24769)
21 (balance* adj2 electrolyte* adj2 solution*).tw. (201)
22 fluid*.tw. (390102)
23 oral* rehydrat*.tw. (2680)
24 (Hartmann* or PlasmaLyte).tw. (3287)
25 or/19–24 (425852)
26 Economics/ (27020)
27 exp “Costs and Cost Analysis”/ (223430)
28 Economics, Dental/ (1902)
29 exp Economics, Hospital/ (23455)
30 exp Economics, Medical/ (14090)
31 Economics, Nursing/ (3986)
32 Economics, Pharmaceutical/ (2855)
33 Budgets/ (11084)
34 exp Models, Economic/ (13983)
35 Markov Chains/ (13310)
36 Monte Carlo Method/ (26563)
37 Decision Trees/ (10506)
38 econom$.tw. (216611)
39 cba.tw. (9519)
40 cea.tw. (19500)
41 cua.tw. (927)
42 markov$.tw. (16463)
43 (monte adj carlo).tw. (27925)
44 (decision adj3 (tree$ or analys$)).tw. (11859)
45 (cost or costs or costing$ or costly or costed).tw. (420258)
46 (price$ or pricing$).tw. (30693)
47 budget$.tw. (22163)
48 expenditure$.tw. (45624)
49 (value adj3 (money or monetary)).tw. (1909)
50 (pharmacoeconomic$ or (pharmaco adj economic$)).tw. (3332)
51 or/26–50 (855405)
52 “Quality of Life”/ (174050) 53 quality of life.tw. (205190)
54 “Value of Life”/ (5642)
55 Quality-Adjusted Life Years/ (10861)
56 quality adjusted life.tw. (9490)
57 (qaly$ or qald$ or qale$ or qtime$).tw. (7804)
58 disability adjusted life.tw. (2296)
59 daly$.tw. (2120)
60 Health Status Indicators/ (22807)
61 (sf36 or sf 36 or short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six).tw. (20770)
62 (sf6 or sf 6 or short form 6 or shortform 6 or sf six or sfsix or shortform six or short form six).tw. (1240)
63 (sf12 or sf 12 or short form 12 or shortform 12 or sf twelve or sftwelve or shortform twelve or short form twelve).tw. (4349)
64 (sf16 or sf 16 or short form 16 or shortform 16 or sf sixteen or sfsixteen or shortform sixteen or short form sixteen).tw. (28)
65 (sf20 or sf 20 or short form 20 or shortform 20 or sf twenty or sftwenty or shortform twenty or short form twenty).tw. (368)
66 (euroqol or euro qol or eq5d or eq 5d).tw. (7484)
67 (qol or hql or hqol or hrqol).tw. (38930)
68 (hye or hyes).tw. (58)
69 health$ year$ equivalent$.tw. (38)
70 utilit$.tw. (155766)
71 (hui or hui1 or hui2 or hui3).tw. (1179)
72 disutili$.tw. (340)
73 rosser.tw. (82)
74 quality of wellbeing.tw. (11)
75 quality of well-being.tw. (366)
76 qwb.tw. (186)
77 willingness to pay.tw. (3818)
78 standard gamble$.tw. (752)
79 time trade off.tw. (966)
80 time tradeoff.tw. (223)
81 tto.tw. (829)
82 or/52–81 (446712)
83 51 or 82 (1240281)
84 and/10,18,83 (96)
85 (201301* or 201302* or 201303* or 201304* or 201305* or 201306* or 201307* or 201308* or 201309* or 201310* or 201311* or 201312* or 2014* or 2015* or 2016* or 2017* or 2018* or 2019*).ed. (5212453)
86 84 and 85 (26)
87 and/10,25,83 (43)
88 86 or 87 (60)
89 limit 88 to english language (51)
90 animals/ not humans/ (4533951)
91 89 not 90 (50)

Database: MiP/Epubs
Strategy used:
1 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*).tw. (135)
2 (contrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (15)
3 (radiocontrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (1)
4 (contrast* adj4 (nephropath* or nephrotoxi*)).tw. (40)
5 (radiocontrast* adj4 (nephropath* or nephrotoxi*)).tw. (3)
6 or/1–5 (158)
7 Saline*.tw. (2003)
8 (Nacet* or acet* or n-acet* or parvolex or mucomyst).tw. (7182)
9 (sodium adj2 (chloride* or bicarbonat*)).tw. (297)
10 or/7–9 (9357)
11 (balance* adj2 electrolyte* adj2 solution*).tw. (2)
12 fluid*.tw. (7568)
13 oral* rehydrat*.tw. (17)
14 (Hartmann* or PlasmaLyte).tw. (58)
15 or/11–14 (7635)
16 econom*.tw. (6178)
17 cba.tw. (56)
18 cea.tw. (331)
19 cua.tw. (22)
20 markov*.tw. (816)
21 (monte adj carlo).tw. (2255)
22 (decision adj3 (tree* or analys*)).tw. (360)
23 (cost or costs or costing* or costly or costed).tw. (12485)
24 (price* or pricing*).tw. (880)
25 budget*.tw. (591)
26 expenditure*.tw. (1193)
27 (value adj3 (money or monetary)).tw. (76)
28 (pharmacoeconomic* or (pharmaco adj economic*)).tw. (51)
29 or/16–28 (21860)
30 quality of life.tw. (6434)
31 quality adjusted life.tw. (332)
32 (qaly* or qald* or qale* or qtime*).tw. (283)
33 disability adjusted life.tw. (91)
34 daly*.tw. (84)
35 (sf36 or sf 36 or short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six).tw. (449)
36 (sf6 or sf 6 or short form 6 or shortform 6 or sf six or sfsix or shortform six or short form six).tw. (79)
37 (sf12 or sf 12 or short form 12 or shortform 12 or sf twelve or sftwelve or shortform twelve or short form twelve).tw. (135)
38 (sf16 or sf 16 or short form 16 or shortform 16 or sf sixteen or sfsixteen or shortform sixteen or short form sixteen).tw. (0)
39 (sf20 or sf 20 or short form 20 or shortform 20 or sf twenty or sftwenty or shortform twenty or short form twenty).tw. (6)
40 (euroqol or euro qol or eq5d or eq 5d).tw. (332)
41 (qol or hql or hqol or hrqol).tw. (1270)
42 (hye or hyes).tw. (2)
43 health* year* equivalent*.tw. (0)
44 utilit*.tw. (4921)
45 (hui or hui1 or hui2 or hui3).tw. (23)
46 disutili*.tw. (18)
47 rosser.tw. (0)
48 quality of wellbeing.tw. (1)
49 quality of well-being.tw. (6)
50 qwb.tw. (3)
51 willingness to pay.tw. (141)
52 standard gamble*.tw. (9)
53 time trade off.tw. (30)
54 time tradeoff.tw. (5)
55 tto.tw. (21)
56 or/30–55 (11663)
57 29 or 56 (31881)
58 and/6,10,57 (2)
59 (201301* or 201302* or 201303* or 201304* or 201305* or 201306* or 201307* or 201308* or 201309* or 201310* or 201311* or 201312* or 2014* or 2015* or 2016* or 2017* or 2018* or 2019*).dt. (282791)
60 58 and 59 (2)
61 and/6,15,57 (0)
62 60 or 61 (2)
63 limit 62 to english language (2)

Database: Embase
Strategy used:
1 exp acute kidney failure/ (71966)
2 exp acute kidney tubule necrosis/ (4531)
3 1 or 2 (75020)
4 (contrast-induced* or radiocontrast-induced* or ci).tw. (746179)
5 3 and 4 (6655)
6 contrast medium/ae (5571)
7 contrast induced nephropathy/ (4314)
8 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*).tw. (14398)
9 (contrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (1045)
10 (radiocontrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw. (60)
11 (radiocontrast* adj4 (nephropath* or nephrotoxi*)).tw. (357)
12 or/5–11 (27080)
13 acetylcysteine/ (33716)
14 sodium chloride/ (171848)
15 bicarbonate/ (45173)
16 hypertonic solution/ (3912)
17 Saline*.tw. (222699)
18 (Nacet* or acet* or n-acet* or parvolex or mucomyst).tw. (693221)
19 (sodium adj2 (chloride* or bicarbonat*)).tw. (30535)
20 or/13–19 (1044216)
21 exp infusion fluid/ (29634)
22 oral rehydration therapy/ (2559)
23 electrolyte/ (36221)
24 acetic acid plus gluconate sodium plus magnesium chloride plus potassium chloride plus sodium chloride/ (312)
25 (balance* adj2 electrolyte* adj2 solution*).tw. (254)
26 fluid*.tw. (544967)
27 oral* rehydrat*.tw. (3031)
28 (Hartmann* or PlasmaLyte).tw. (5022)
29 or/21–28 (605613)
30 exp Health Economics/ (788191)
31 exp “Health Care Cost”/ (272876)
32 exp Pharmacoeconomics/ (191839)
33 Monte Carlo Method/ (35445)
34 Decision Tree/ (10807)
35 econom$.tw. (327324)
36 cba.tw. (12180)
37 cea.tw. (31923)
38 cua.tw. (1343)
39 markov$.tw. (26602)
40 (monte adj carlo).tw. (42303)
41 (decision adj3 (tree$ or analys$)).tw. (20041)
42 (cost or costs or costing$ or costly or costed).tw. (683195)
43 (price$ or pricing$).tw. (51316)
44 budget$.tw. (35046)
45 expenditure$.tw. (67959)
46 (value adj3 (money or monetary)).tw. (3110)
47 (pharmacoeconomic$ or (pharmaco adj economic$)).tw. (8163)
48 or/30–47 (1590045)
49 “Quality of Life”/ (418371)
50 Quality Adjusted Life Year/ (23273)
51 Quality of Life Index/ (2585)
52 Short Form 36/ (25191)
53 Health Status/ (118404)
54 quality of life.tw. (384950)
55 quality adjusted life.tw. (17080)
56 (qaly$ or qald$ or qale$ or qtime$).tw. (17531)
57 disability adjusted life.tw. (3412)
58 daly$.tw. (3390)
59 (sf36 or sf 36 or short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six).tw. (38030)
60 (sf6 or sf 6 or short form 6 or shortform 6 or sf six or sfsix or shortform six or short form six).tw. (2121)
61 (sf12 or sf 12 or short form 12 or shortform 12 or sf twelve or sftwelve or shortform twelve or short form twelve).tw. (8388)
62 (sf16 or sf 16 or short form 16 or shortform 16 or sf sixteen or sfsixteen or shortform sixteen or short form sixteen).tw. (54)
63 (sf20 or sf 20 or short form 20 or shortform 20 or sf twenty or sftwenty or shortform twenty or short form twenty).tw. (416)
64 (euroqol or euro qol or eq5d or eq 5d).tw. (17336)
65 (qol or hql or hqol or hrqol).tw. (84264)
66 (hye or hyes).tw. (122)
67 health$ year$ equivalent$.tw. (40)
68 utilit$.tw. (256523)
69 (hui or hui1 or hui2 or hui3).tw. (2029)
70 disutili$.tw. (813)
71 rosser.tw. (110)
72 quality of wellbeing.tw. (38)
73 quality of well-being.tw. (464)
74 qwb.tw. (234)
75 willingness to pay.tw. (7367)
76 standard gamble$.tw. (1045)
77 time trade off.tw. (1578)
78 time tradeoff.tw. (269)
79 tto.tw. (1499)
80 or/49–79 (880430)
81 48 or 80 (2330456)
82 and/12,20,81 (263)
83 (201301* or 201302* or 201303* or 201304* or 201305* or 201306* or 201307* or 201308* or 201309* or 201310* or 201311* or 201312* or 2014* or 2015* or 2016* or 2017* or 2018* or 2019*).dc. (9825657)
84 82 and 83 (109)
85 and/12,29,81 (81)
86 84 or 85 (179)
87 limit 86 to english language (173)
88 limit 87 to (conference abstract or conference paper) (57)
89 87 not 88 (116)
90 nonhuman/ not (human/ and nonhuman/) (4339559)
91 89 not 90 (115)

Database: Econlit
Strategy used:
1 health*.sh,kw. (52846)
2 related disciplines.sh,kw. (8641)
3 1 or 2 (61487)
4 (contrast-induced* or radiocontrast-induced* or ci).tw,kw. (536)
5 3 and 4 (209)
6 (ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*).tw,kw. (12)
7 (contrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw,kw. (0)
8 (radiocontrast-induced adj4 (aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*)).tw,kw. (0)
9 (contrast* adj4 (nephropath* or nephrotoxi*)).tw,kw. (1)
10 (radiocontrast* adj4 (nephropath* or nephrotoxi*)).tw,kw. (0)
11 or/5–10 (221)
12 Saline*.tw,kw. (71)
13 (Nacet* or acet* or n-acet* or parvolex or mucomyst).tw,kw. (42)
14 (sodium adj2 (chloride* or bicarbonat*)).tw,kw. (0)
15 or/12–14 (113)
16 (balance* adj2 electrolyte* adj2 solution*).tw,kw. (0)
17 fluid*.tw,kw. (834)
18 oral* rehydrat*.tw,kw. (18)
19 (Hartmann* or PlasmaLyte).tw,kw. (34)
20 or/16–19 (885)
21 11 and 15 (0)
22 11 and 20 (3)
23 21 or 22 (3)

Database: NHS EED
Strategy used:
1 MeSH DESCRIPTOR acute kidney injury EXPLODE ALL TREES
2 ((contrast-induced* or radiocontrast-induced* or ci))
3 #1 AND #2
4 MeSH DESCRIPTOR contrast media WITH QUALIFIER AE
5 ((ciaki or cin or ciraf or ci-aki or ci-arf or ci nephropath* or cinephropath* or rci nephropath* or rcinephropath*))
6 (contrast-induced) AND ((aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*))
7 (radiocontrast-induced) AND ((aki or arf or acute kidney or acute renal or early kidney or early renal or necrosis or tubul*))
8 (contrast) AND ((nephropath* or nephrotoxi*))
9 (radiocontrast) AND ((nephropath* or nephrotoxi*))
10 #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9
11 MeSH DESCRIPTOR Acetylcysteine
12 MeSH DESCRIPTOR Sodium Chloride
13 MeSH DESCRIPTOR Bicarbonates
14 MeSH DESCRIPTOR Saline Solution, Hypertonic
15 (Saline*)
16 ((Nacet* or acet* or n-acet* or parvolex or mucomyst))
17 (Sodium) AND ((chloride* or bicarbonat*))
18 #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17
19 MeSH DESCRIPTOR Fluid therapy EXPLODE ALL TREES
20 MeSH DESCRIPTOR Electrolytes
21 (balance) AND (electrolyte) AND (solution*)
22 (fluid*)
23 (oral* rehydrat*)
24 ((Hartmann* or PlasmaLyte))
25 #19 OR #20 OR #21 OR #22 OR #23 OR #24
26 #10 AND #18
27 (#26) IN NHSEED FROM 2013 TO 2019
28 #10 AND #25
29 (#28) IN NHSEED
30 #27 OR #29

Appendix D. Clinical evidence study selection

Appendix E. Clinical evidence tables

Download PDF (1.5M)

Appendix F. Forest plots of pairwise meta-analysis

Appendix G. Network meta-analysis results

Model fit statistics

Table 11. Model fit statistics

Network diagram

Figure 31. The thickness of the line represents the number of studies

Caterpillar plot

Figure 32. Relative effectiveness of all options versus no (intravenous) hydration. (Mean differences with 95% credible intervals and line of no effect in red; values higher than 1.0 favour no (intravenous) hydration; values lower than 1.0 favour the other treatments.)

Rank probability histograms

Figure 33

Probability of the treatment assuming each treatment rank (see treatment codes below;rank 1 is best).

Relative effectiveness

Table 12. Relative effectiveness of all pairwise combinations. (Upper diagonal: risk ratios (RR) with 95% confidence intervals from the pair-wise meta-analysis. RRs less than 1 favour the column defining treatment, RRs greater than 1 favour the row defining treatment. Lower diagonal: posterior median RRs with 95% credible intervals from NMA results, RR less than 1 favour the row defining treatment. RRs greater than 1 favour the column defining treatment.)

Appendix H. GRADE tables

Pair-wise meta-analysis

Sodium chloride 0.45% vs no (intravenous) hydration

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.45

No (intravenous) hydration

Relative (95% CI)

Absolute

CI-AKI (as reported by study) - sCr <132.6μmol/l

randomised trials¹

very serious²

no serious inconsistency

no serious indirectness

very serious³

none

22/330

(6.7%)

23/330

(7%)

RR 0.96 (0.54 to 1.68)

0 fewer per 100 (from 3 fewer to 5 more)

VERY LOW

1: Chen 2008
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Sodium chloride 0.9% vs no (intravenous) hydration

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9%

No (intravenous) hydration

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

serious³

none

42/446

(9.4%)

49/457

(10.7%)

RR 0.86 (0.6 to 1.24)

2 fewer per 100 (from 4 fewer to 3 more)

LOW

1: Maioli 2011; Nijssen 2017
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses one end of a defined MID interval

Subgroup analyses (outcome: CI-AKI): Sodium chloride 0.9% vs no (intravenous) hydration

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9%	No (intravenous) hydration (diabetes)	Relative (95% CI)	Absolute	Quality
Subgroup: diabetes
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	11/31 (35.5%)	10/34 (29.4%)	RR 1.21 (0.6 to 2.44)	6 more per 100 (from 12 fewer to 42 more)	VERY LOW
Subgroup: older people >75 years
1	randomised trials⁴	serious²	no serious inconsistency	no serious indirectness	very serious³	none	15/36 (41.7%)	11/29 (37.9%)	RR 1.10 (0.6 to 2.01)	4 more per 100 (from 15 fewer to 38 more)	VERY LOW

1: Maioli 2011
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Maioli 2011 reported older people ≥75 years

Other outcomes: Sodium chloride 0.9% vs no (intravenous) hydration

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9%	No (intravenous) hydration	Relative (95% CI)	Absolute	Quality
Mortality: in-hospital mortality
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	5/150 (3.3%)	8/150 (5.3%)	RR 0.62 (0.21 to 1.87)	2 fewer per 100 (from 4 fewer to 5 more)	VERY LOW
Mortality: all-cause mortality
1	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	0/328 (0%)	3/332 (0.9%)	RR 0.14 (0.01 to 2.79)	1 fewer per 100 (from 1 fewer to 2 more)	LOW
Need for renal replacement therapy: dialysis
6	randomised trials⁵	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	28/2441 (1.1%)	27/2468 (1.1%)	RR 1.04 (0.62 to 1.75)	0 more per 100 (from 0 fewer to 1 more)	LOW
Adverse events
2	randomised trials¹^,⁶	serious²	very serious⁷	no serious indirectness	very serious³	none	32/478 (6.7%)	15/482 (3.1%)	RR 4.59 (0.16 to 134.39)	11 more per 100 (from 3 fewer to 100 more)	VERY LOW

1: Maioli 2011
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Nijssen 2017; all-cause mortality within 35 days post-contrast
5: Akyuz 2014; Brar 2008; Masuda 2007; Mueller 2002; Solomon 2015; Weisbord 2018
6: Maioli 2011 (major adverse cardiovascular events: death, recurrent myocardial infarction, repeated urgent PCI, stroke and major bleeding); Nijssen 2017 (symptomatic heart failure)
7: i-squared >66.7%; random effects model was used to account for heterogeneity

Sodium chloride 0.9% vs oral fluids

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9%

Oral fluids

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

17/188

(9%)

11/188

(5.9%)

RR 1.52 (0.72 to 3.2)

3 more per 100 (from 2 fewer to 13 more)

VERY LOW

1: Akyuz 2014; Cho 2010; Wrobel 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: Sodium chloride 0.9% vs oral fluids

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9%	Oral fluids	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/109 (0.92%)	0/116 (0%)	RR 3.19 (0.13 to 77.5)	-	VERY LOW
Need for renal replacement therapy: dialysis
1	randomised trials⁴	serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/109 (0.92%)	0/116 (0%)	RR 3.19 (0.13 to 77.5)	-	VERY LOW
Length of hospital stay in days (Better indicated by lower values)
1	randomised trials⁵	serious²	no serious inconsistency	no serious indirectness	very serious³	none	27	22	-	MD 0.38 lower (3.81 lower to 3.05 higher)	VERY LOW

1: Akyuz 2014; mortality at 30 days
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Akyuz 2014
5: Cho 2010

Sensitivity analysis excluding studies with a high risk of bias: Sodium chloride 0.9% vs oral fluids

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9%

Oral fluids

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

serious³

no serious indirectness

very serious⁴

none

14/136

(10.3%)

9/138

(6.5%)

RR 1.54 (0.68 to 3.51)

4 more per 100 (from 2 fewer to 16 more)

VERY LOW

1: Akyuz 2014; Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: i-squared >33.3%
4: 95% confidence interval crosses both ends of a defined MID interval

Sodium chloride 0.9% vs sodium chloride 0.45%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9%

Sodium chloride 0.45%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

very serious²

no serious inconsistency

no serious indirectness

serious³

none

5/685

(0.73%)

14/698

(2%)

RR 0.36 (0.13 to 1)

1 fewer per 100 (from 2 fewer to 0 more)

VERY LOW

1: Mueller 2002
2: 33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses one end of a defined MID interval

Subgroup analyses (outcome: CI-AKI): Sodium chloride 0.9% vs sodium chloride 0.45%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9%	Sodium chloride 0.45%	Relative (95% CI)	Absolute	Quality
Subgroup: chronic kidney disease
1	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	3/138 (2.2%)	6/148 (4.1%)	RR 0.54 (0.14 to 2.10)	2 fewer per 100 (from 3 fewer to 4 more)	VERY LOW
Subgroup: diabetes
1	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	0/107 (0%)	6/110 (5.5%)	RR 0.08 (0.0 to 1.39)	5 fewer per 100 (from 5 fewer to 2 more)	VERY LOW
Subgroup: low volume of contrast agent
1	randomised trials⁴	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	5/434 (1.2%)	6/430 (1.4%)	RR 0.83 (0.25 to 2.69)	0 fewer per 100 (from 1 fewer to 2 more)	VERY LOW
Subgroup: high volume of contrast agent
1	randomised trials⁵	very serious²	no serious inconsistency	no serious indirectness	serious⁶	none	0/251 (0%)	8/268 (3%)	RR 0.06 (0.0 to 1.08)	28 fewer per 1000 (from 30 fewer to 2 more)	VERY LOW

1: Mueller 2002
2: 33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Mueller 2002 reported low volume of contrast agent <250ml
5: Mueller reported high volume of contrast agent ≥250 ml
6: 95% confidence interval crosses one end of a defined MID interval

Other outcomes: Sodium chloride 0.9% vs sodium chloride 0.45%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9%	Sodium chloride 0.45%	Relative (95% CI)	Absolute	Quality
Mortality
1	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/265 (0.38%)	3/265 (1.1%)	RR 0.33 (0.03 to 3.18)	1 fewer per 100 (from 1 fewer to 2 more)	VERY LOW
Need for renal replacement therapy: dialysis
1	randomised trials⁴	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/685 (0.15%)	1/698 (0.14%)	RR 1.02 (0.06 to 16.26)	0 more per 100 (from 0 fewer to 2 more)	VERY LOW
Adverse events
1	randomised trials⁵	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	14/265 (5.3%)	17/265 (6.4%)	RR 0.82 (0.41 to 1.64)	1 fewer per 100 (from 4 fewer to 4 more)	VERY LOW

1: Mueller 2002; mortality within 30 days in subgroup of people receiving coronary artery stents
2: 33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Mueller 2002
5: Mueller 2002; major adverse cardiac events within 30 days in a predefined subgroup of 530 patients receiving coronary artery stents

Sodium chloride 0.9% vs sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9%

Sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials

no serious risk of bias

serious¹

no serious indirectness

no serious imprecision

none

262/2716

(9.6%)

249/2696

(9.2%)

RR 1.04 (0.88 to 1.23)

0 more per 100 (from 1 fewer to 2 more)

MODERATE

1: Adolph 2008; Boucek 2013; Brar 2008; Castini 2010; Cho 2010; Hafiz 2012; Kooiman 2014a; Kooiman 2018; Maioli 2011; Masuda 2007; Merten 2004; Nieto-Rios 2014; Solomon 2015; van Mourik 2018; Weisbord 2018
2: i-squared >33.3%

Other outcomes: Sodium chloride 0.9% vs sodium bicarbonate

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9%	Sodium bicarbonate	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (30 days)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	3/178 (1.7%)	3/175 (1.7%)	RR 0.98 (0.2 to 4.8)	0 fewer per 100 (from 1 fewer to 7 more)	VERY LOW
Mortality: all-cause mortality (>30 days)
3	randomised trials⁴	no serious risk of bias	serious⁵	no serious indirectness	very serious³	none	51/1618 (3.2%)	44/1624 (2.7%)	RR 1.36 (0.65 to 2.83)	1 more per 100 (from 1 fewer to 5 more)	VERY LOW
Mortality: in-hospital mortality
2	randomised trials⁶	serious²	no serious inconsistency	no serious indirectness	very serious³	none	7/179 (3.9%)	3/180 (1.7%)	RR 2.05 (0.57 to 7.35)	2 more per 100 (from 1 fewer to 11 more)	VERY LOW
Need for renal replacement therapy
4	randomised trials⁷	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	24/1647 (1.5%)	26/1654 (1.6%)	RR 0.93 (0.54 to 1.61)	0 fewer per 100 (from 1 fewer to 1 more)	LOW
Adverse events
3	randomised trials⁸	serious²	no serious inconsistency	no serious indirectness	serious⁹	none	26/387 (6.7%)	15/386 (3.9%)	RR 1.74 (0.94 to 3.21)	3 more per 100 (from 0 fewer to 9 more)	LOW
Adverse events: heart failure
3	randomised trials¹⁰	very serious¹¹	serious⁵	no serious indirectness	very serious³	none	24/431 (5.6%)	14/414 (3.4%)	RR 1.80 (0.59 to 5.48)	3 more per 100 (from 1 fewer to 15 more)	VERY LOW
Length of hospital stay in days (Better indicated by lower values)
2	randomised trials¹²	serious²	no serious inconsistency	no serious indirectness	very serious³	none	91	83	-	MD 0.06 lower (2.3 lower to 2.18 higher)	VERY LOW

1: Brar 2008
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Brar 2008 (30 days to 6 months); Solomon 2015 (6 months): Weisbord 2018 (3 months)
5: i-squared >33.3%
6: Maioli 2011; Masuda 2007
7: Brar 2008 (dialysis); Masuda 2007 (hemodialysis); Solomon 2015 (dialysis); Weisbord 2018 (dialysis)
8: Boucek 2013 (any adverse events); Brar 2008 (mycardial infarction, cerebrovascular accident [stroke and transient ischemic attack]); Maioli 2011 (major adverse cardiovascular events: death, recurrent myocardial infarction, repeated urgent PCI, stroke & major bleeding)
9: 95% confidence interval crosses one end of a defined MID interval
10: Kooiman 2014 (acute heart failure); Masuda 2007 (heart failure); Nieto-Rios 2014 (decompensated heart failure)
11: >33.3% of weighted data from studies at high risk of bias
12: Boucek 2013; Cho 2010

Sensitivity analysis excluding studies with a high risk of bias: Sodium chloride 0.9% vs sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9%

Sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

no serious imprecision

none

252/2687

(9.4%)

247/2666

(9.3%)

RR 1.01 (0.85 to 1.19)

0 more per 100 (from 1 fewer to 2 more)

HIGH

1: Adolph 2008; Boucek 2013; Brar 2008; Castini 2010; Cho 2010; Hafiz 2012; Kooiman 2014a; Kooiman 2018; Maioli 2011; Merten 2004; Nieto-Rios 2014; Solomon 2015; van Mourik 2018; Weisbord 2018

Sodium chloride 0.9% vs oral sodium bicarbonate + oral fluids

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV sodium chloride 0.9%

Oral sodium bicarbonate + oral fluids

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

6/27

(22.2%)

1/21

(4.8%)

RR 4.67 (0.61 to 35.84)

17 more per 100 (from 2 fewer to 100 more)

VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: Sodium chloride 0.9% vs oral sodium bicarbonate + oral fluids

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9%	Oral sodium bicarbonate + oral fluids	Relative (95% CI)	Absolute	Quality
Length of hospital stay in days (Better indicated by lower values)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	27	21	-	MD 2.72 lower (7.25 lower to 1.81 higher)	VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Sodium chloride 0.9% (5 hours) vs sodium chloride 0.9% (20 hours)

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9% (5 hours)

Sodium chloride 0.9% (20 hours)

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

2/60

(3.3%)

2/62

(3.2%)

RR 1.03 (0.15 to 7.1)

0 more per 100 (from 3 fewer to 20 more)

VERY LOW

1: Torigoe 2013
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Sodium chloride 0.45% + sodium bicarbonate vs sodium chloride 0.45%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.45% + sodium bicarbonate

Sodium chloride 0.45%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

5/36

(13.9%)

4/36

(11.1%)

RR 1.25 (0.37 to 4.28)

3 more per 100 (from 7 fewer to 36 more)

VERY LOW

1: Vasheghani-Farahani 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Sodium chloride 0.9% + sodium bicarbonate vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9% + sodium bicarbonate

Sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

serious³

no serious indirectness

serious⁴

none

25/271

(9.2%)

47/271

(17.3%)

RR 0.46 (0.19 to 1.11)

9 fewer per 100 (from 14 fewer to 2 more)

VERY LOW

1: Kama 2014; Motohiro 2011; Tamura 2009; Turedi 2016
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: i-squared >33.3%
4: 95% confidence interval crosses one end of a defined MID interval

Other outcomes: Sodium chloride 0.9% + sodium bicarbonate vs sodium chloride 0.9%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9% + sodium bicarbonate	Sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Mortality: in-hospital mortality
1	randomised trials¹	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	10/85 (11.8%)	12/87 (13.8%)	RR 0.85 (0.39 to 1.87)	2 fewer per 100 (from 8 fewer to 12 more)	LOW
Need for renal replacement therapy
3	randomised trials³	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	11/193 (5.7%)	16/194 (8.2%)	RR 0.72 (0.36 to 1.44)	2 fewer per 100 (from 5 fewer to 4 more)	LOW
Adverse events
1	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	0/72 (0%)	1/72 (1.4%)	RR 0.33 (0.01 to 8.05)	1 fewer per 100 (from 1 fewer to 10 more)	LOW

1: Turedi 2016
2: 95% confidence interval crosses both ends of a defined MID interval
3: Kama 2014 (type of RRT was not reported); Tamura 2009 (hemodialysis); Turedi 2016 (hemodialysis or peritoneal dialysis)
4: Tamura 2009 (adverse clinical events within first 7 days after procedure)

Sodium chloride 0.9% + sodium bicarbonate vs sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium chloride 0.9% + sodium bicarbonate

Sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

serious³

none

8/29

(27.6%)

2/30

(6.7%)

RR 4.14 (0.96 to 17.87)

21 more per 100 (from 0 fewer to 100 more)

LOW

1: Ueda 2011
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses one end of a defined MID interval

Other outcomes: Sodium chloride 0.9% + sodium bicarbonate vs sodium bicarbonate

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium chloride 0.9% + sodium bicarbonate	Sodium bicarbonate	Relative (95% CI)	Absolute	Quality
Mortality: in-hospital mortality
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	2/29 (6.9%)	3/30 (10%)	RR 0.69 (0.12 to 3.83)	3 fewer per 100 (from 9 fewer to 28 more)	VERY LOW
Adverse events
1	randomised trials¹^,⁴	serious²	no serious inconsistency	no serious indirectness	very serious³	none	10/29 (34.5%)	9/30 (30%)	RR 1.15 (0.55 to 2.41)	4 more per 100 (from 14 fewer to 42 more)	VERY LOW
Length of hospital stay in days (Better indicated by lower values)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	30	30	-	MD 1.40 lower (10.90 lower to 8.10 higher)	VERY LOW

1: Ueda 2011
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Overall adverse events: congestive heart failure, acute renal failure requiring dialysis, lethal arrythmia and death

Oral sodium bicarbonate + oral fluids vs oral fluids

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral sodium bicarbonate + oral fluids

Oral fluids

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

1/21

(4.8%)

1/22

(4.5%)

RR 1.05 (0.07 to 15.69)

0 more per 100 (from 4 fewer to 67 more)

VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: Oral sodium bicarbonate + oral fluids vs oral fluids

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral sodium bicarbonate + oral fluids	Oral fluids	Relative (95% CI)	Absolute	Quality
Length of hospital stay (Better indicated by lower values)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	21	22	-	MD 2.54 higher (2.32 lower to 7.40 higher)	VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Sodium bicarbonate vs no (intravenous) hydration

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium bicarbonate

No (intravenuos) hydration

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

no serious imprecision

none

26/263

(9.9%)

51/259

(19.7%)

RR 0.51 (0.33 to 0.78)

10 fewer per 100 (from 4 fewer to 13 fewer)

MODERATE

1: Kooiman 2014b; Maioli 2011; Martin-Moreno 2015
2: >33.3% of weighted data from studies at moderate or high risk of bias

Subgroup analyses (outcome: CI-AKI): Sodium bicarbonate vs no (intravenous) hydration

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium bicarbonate	No (intravenous) hydration	Relative (95% CI)	Absolute	Quality
Subgroup: diabetes
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	5/31 (16.1%)	10/34 (29.4%)	RR 0.55 (0.21 to 1.43)	13 fewer per 100 (from 23 fewer to 13 more)	VERY LOW
Subgroup: older people >75 years
1	randomised trials⁴	serious²	no serious inconsistency	no serious indirectness	serious⁵	none	8/38 (21.1%)	11/29 (37.9%)	RR 0.56 (0.26 to 1.2)	17 fewer per 100 (from 28 fewer to 8 more)	LOW

1: Maioli 2011
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Maioli 2011 reported older people ≥75 years
5: 95% confidence interval crosses one end of a defined MID interval

Sensitivity analysis excluding studies with a high risk of bias: Sodium bicarbonate vs no (intravenous) hydration

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium bicarbonate

No (intravenous) hydration

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

no serious imprecision

none

23/220

(10.5%)

47/215

(21.9%)

RR 0.48 (0.31 to 0.77)

11 fewer per 100 (from 5 fewer to 15 fewer)

MODERATE

1: Kooiman 2014b; Maioli 2011
2: >33.3% of weighted data from studies at moderate or high risk of bias

Sodium bicarbonate vs oral fluids

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium bicarbonate

Oral fluids

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

2/21

(9.5%)

1/22

(4.5%)

RR 2.1 (0.2 to 21.42)

5 more per 100 (from 4 fewer to 93 more)

VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: Sodium bicarbonate vs oral fluids

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium bicarbonate	Oral fluids	Relative (95% CI)	Absolute	Quality
Length of hospital stay in days (Better indicated by lower values)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	21	22	-	MD 0.27 lower (3.48 lower to 2.94 higher)	VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Sodium bicarbonate vs oral sodium citrate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Sodium bicarbonate

Oral sodium citrate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

very serious²

no serious inconsistency

no serious indirectness

very serious³

none

3/43

(7%)

5/43

(11.6%)

RR 0.6 (0.15 to 2.36)

5 fewer per 100 (from 10 fewer to 16 more)

VERY LOW

1: Martin-Moreno 2015
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Sodium bicarbonate vs oral sodium bicarbonate + oral fluids

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV sodium bicarbonate

Oral sodium bicarbonate + oral fluids

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

2/21

(9.5%)

1/21

(4.8%)

RR 2 (0.2 to 20.41)

5 more per 100 (from 4 fewer to 92 more)

VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: Sodium bicarbonate vs oral sodium bicarbonate + oral fluids

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Sodium bicarbonate	Oral sodium bicarbonate + oral fluids	Relative (95% CI)	Absolute	Quality
Length of hospital stay in days (Better indicated by lower values)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	27	21	-	MD 2.81 lower (7.10 lower to 1.48 higher)	VERY LOW

1: Cho 2010
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Oral sodium citrate vs no (intravenous) hydration

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral sodium citrate

No (intravenuos) hydration

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

very serious²

no serious inconsistency

no serious indirectness

very serious³

none

5/43

(11.6%)

4/44

(9.1%)

RR 1.28 (0.37 to 4.45)

3 more per 100 (from 6 fewer to 31 more)

VERY LOW

1: Martin-Moreno 2015
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Oral NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.45%

Sodium chloride 0.45

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

very serious²

serious³

no serious indirectness

no serious imprecision

none

49/583

(8.4%)

96/571

(16.8%)

RR 0.50 (0.37 to 0.70)

8 fewer per 100 (from 5 fewer to 11 fewer)

VERY LOW

1: Agrawal 2004; Allaqaband 2002; Briguori 2002; Durham 2002; Goldenberg 2004; Hsu 2007; Izani Wan Mohamed 2008; Kitzler 2012; MacNeill 2003; Miner 2004; Oldemeyer 2003; Seyon 2007; Shyu 2002; Tepel 2000
2: >33.3% of weighted data from studies at high risk of bias
3: i-squared >33.3%

Subgroup analyses (outcome: CI-AKI): Oral NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.45%	Sodium chloride 0.45% (CKD)	Relative (95% CI)	Absolute	Quality
Subgroup: chronic kidney disease
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	3/7 (42.9%)	5/12 (41.7%)	RR 1.03 (0.35 to 3.05)	1 more per 100 (from 27 fewer to 85 more)	VERY LOW
Subgroup: diabetes
2	randomised trials⁴	serious²	no serious inconsistency	no serious indirectness	very serious³	none	13/64 (20.3%)	8/58 (13.8%)	RR 1.5 (0.7 to 3.24)	7 more per 100 (from 4 fewer to 31 more)	VERY LOW

1: Durham 2002
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Durham 2002, Allaqaband 2002

Other outcomes: Oral NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.45%	Sodium chloride 0.45	Relative (95% CI)	Absolute	Quality
Mortality: long-term mortality
1	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	4/95 (4.2%)	3/85 (3.5%)	RR 1.19 (0.27 to 5.18)	1 more per 100 (from 3 fewer to 15 more)	VERY LOW
Mortality: in-hospital mortality
1	randomised trials⁴	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	0/95 (0%)	2/85 (2.4%)	RR 0.18 (0.01 to 3.68)	2 fewer per 100 (from 2 fewer to 6 more)	VERY LOW
Need for renal replacement therapy: dialysis
3	randomised trials⁵	serious⁶	no serious inconsistency	no serious indirectness	very serious³	none	1/247 (0.4%)	2/237 (0.84%)	RR 0.69 (0.13 to 3.52)	0 fewer per 100 (from 1 fewer to 2 more)	VERY LOW
Adverse events
5	randomised trials⁷	very serious²	no serious inconsistency⁸	no serious indirectness	serious³^,⁹	none	41/329 (12.5%)	23/307 (7.5%)	RR 1.61 (1.01 to 2.56)	5 more per 100 (from 0 more to 12 more)	VERY LOW
Length of hospital stay in days (Better indicated by lower values)
2	randomised trials¹⁰	very serious²	serious⁸	no serious indirectness	very serious³	none	60	56	-	MD 1.24 lower (3.94 lower to 1.45 higher)	VERY LOW
Readmission for AKI
1	randomised trials⁴	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	13/95 (13.7%)	13/85 (15.3%)	RR 0.89 (0.44 to 1.82)	2 fewer per 100 (from 9 fewer to 13 more)	VERY LOW

1: Miner 2004; long-term follow-up but time was not reported
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Miner 2004
5: Briguori 2002; Miner 2004; Shyu 2002
6: >33.3% of weighted data from studies at moderate or high risk of bias
7: Goldenberg 2004 (congestive heart failure); Izani Wan Mahamed 2008 (adverse events); Miner 2004 (in-hospital adverse events and long-term clinical events but time was not reported); Oldemeyer 2003 (adverse events)
8: i-squared >33.3%
9: 95% confidence interval crosses one end of a defined MID interval
10: Hsu 2007; Oldemeyer 2003

Sensitivity analysis excluding studies with a high risk of bias: Oral NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.45%

Sodium chloride 0.45

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

serious³

no serious indirectness

serious⁴

none

39/456

(8.6%)

65/455

(14.3%)

RR 0.61 (0.42 to 0.88)

6 fewer per 100 (from 2 fewer to 8 fewer)

VERY LOW

1: Agrawal 2004; Allaqaband 2002; Briguori 2002; Durham 2002; Goldenberg 2004; Izani Wan Mohamed 2008; Kitzler 2012; Oldemeyer 2003; Seyon 2007; Shyu 2002; Tepel 2000
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: i-squared >33.3%
4: 95% confidence interval crosses one end of a defined MID interval

Oral NAC + sodium chloride 0.45% vs oral NAC

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.45%

Oral NAC

Relative (95% CI)

Absolute

CI-AKI (as reported by study) - sCr ≥132.6μmol/l

randomised trials¹

very serious²

no serious inconsistency

no serious indirectness

serious³

none

40/188

(21.3%)

64/188

(34%)

RR 0.62 (0.45 to 0.88)

13 fewer per 100 (from 4 fewer to 19 fewer)

VERY LOW

1: Chen 2008
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses one end of a defined MID interval

Oral NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.9%

Sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

serious²

none

341/3307

(10.3%)

352/3290

(10.7%)

RR 0.96 (0.83 to 1.10)

0 fewer per 100 (from 2 fewer to 1 more)

MODERATE

1: ACT investigators 2011; Albabtain 2013; Baskurt 2009; Castini 2010; Erturk 2014; Ferrario 2009; Fung 2004; Gomes 2005; Habib 2016; Hafiz 2012; Kay 2003; Khalili 2006; Reinecke 2007; Sadineni 2017; Saitoh; Weisbord 2018
2: 95% confidence interval crosses one end of a defined MID interval

Subgroup analyses (outcome: CI-AKI): Oral NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.9%	Sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Subgroup: chronic kidney disease
1	randomised trials¹	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	12/188 (6.4%)	10/179 (5.6%)	RR 1.14 (0.51 to 2.58)	1 more per 100 (from 3 fewer to 9 more)	LOW
Subgroup: diabetes
5	randomised trials³	no serious risk of bias	no serious inconsistency	no serious indirectness	serious⁴	none	110/797 (13.8%)	113/769 (14.7%)	RR 0.95 (0.75 to 1.21)	1 fewer per 100 (from 4 fewer to 3 more)	MODERATE
Subgroup: older people >75 years
1	randomised trials⁵	very serious⁶	no serious inconsistency	no serious indirectness	very serious²	none	4/7 (57.1%)	8/11 (72.7%)	RR 0.79 (0.38 to 1.64)	15 fewer per 100 (from 45 fewer to 47 more)	VERY LOW
Subgroup: low volume of contrast agent
1	randomised trials⁷	very serious⁶	no serious inconsistency	no serious indirectness	very serious²	none	6/7 (85.7%)	8/11 (72.7%)	RR 1.18 (0.74 to 1.89)	13 more per 100 (from 19 fewer to 65 more)	VERY LOW
Subgroup: high volume of contrast agent
2	randomised trials⁸	very serious⁶	no serious inconsistency	no serious indirectness	very serious²	none	7/66 (10.6%)	7/59 (11.9%)	RR 0.98 (0.35 to 2.72)	0 fewer per 100 (from 8 fewer to 20 more)	VERY LOW

1: ACT investigators 2011
2: 95% confidence interval crosses both ends of a defined MID interval
3: ACT investigators 2011, Ferrario 2009; Fung 2004, Gomes 2005; Sadineni 2017
4: 95% confidence interval crosses one end of a defined MID interval
5: Sadineni 2017 reported older people >60 years
6: >33.3% of weighted data from studies at high risk of bias
7: Sadineni 2017 reported low volume of contrast agent <100ml
8: Ferrario 2008 reported high volume of contrast agent >140 ml; Sadineni 2017 reported high volume of contrast agent ≥100 ml

Other outcomes: Oral NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.9%	Sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (30 days)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	0/102 (0%)	3/103 (2.9%)	RR 0.14 (0.01 to 2.76)	3 fewer per 100 (from 3 fewer to 5 more)	VERY LOW
Mortality: all-cause mortality (30 days - 1 year)
2	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	serious⁵	none	48/1340 (3.6%)	35/1347 (2.6%)	RR 1.38 (0.9 to 2.12)	1 more per 100 (from 0 fewer to 3 more)	MODERATE
In-hospital mortality
1	randomised trials⁶	serious²	no serious inconsistency	no serious indirectness	very serious³	none	5/77 (6.5%)	2/79 (2.5%)	RR 2.56 (0.51 to 12.83)	4 more per 100 (from 1 fewer to 30 more)	VERY LOW
1	randomised trials⁶	serious²	no serious inconsistency	no serious indirectness	very serious³	none		0%	RR 2.56 (0.51 to 12.83)	-	VERY LOW
Need for renal replacement therapy: dialysis
6	randomised trials⁷	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	21/2769 (0.76%)	25/2731 (0.92%)	RR 0.83 (0.48 to 1.46)	0 fewer per 100 (from 0 fewer to 0 more)	LOW
Adverse events
4	randomised trials⁸	no serious risk of bias	no serious inconsistency	no serious indirectness	serious⁵	none	111/2492 (4.5%)	114/2415 (4.7%)	RR 0.94 (0.73 to 1.22)	0 fewer per 100 (from 1 fewer to 1 more)	MODERATE
Hospital length of stay in days (Better indicated by lower values)
1	randomised trials⁹	no serious risk of bias	no serious inconsistency	no serious indirectness	no serious imprecision	none	102	98	-	MD 0.50 lower (0.93 to 0.07 lower)	HIGH

1: Ertuk 2014
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Ertuk 2014 (1 year); Weisbord 2018 (3 months)
5: 95% confidence interval crosses one end of a defined MID interval
6: Gomes 2005
7: ACT investigators 2011; Ertuk 2014; Gomes 2005; Reinecke 2007; Sadineni 2017; Weisbord 2018
8: ACT investigators 2011 (any adverse events or any serious adverse events [stroke, myocardial infarction, pneumonia, sepsis and acute pulmonary edema]); Fung 2004 (clinical heart failure so patients could not complete sodium chloride infusion regimen); Kay 2003 (congestive heart failure, uncomplicated non-ST-segment elevation mycardial infarction, nausea)
9: Kay 2003

Sensitivity analysis excluding studies with a high risk of bias: Oral NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.9%

Sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

no serious imprecision

none

326/3128

(10.4%)

326/3100

(10.5%)

RR 0.99 (0.85 to 1.14)

0 fewer per 100 (from 2 fewer to 1 more)

HIGH

1: ACT investigators 2011; Albabtain 2013; Baskurt 2009; Castini 2010; Erturk 2014; Ferrario 2009; Fung 2004; Gomes 2005; Hafiz 2012; Kay 2003; Khalili 2006; Saitoh 2011; Weisbord 2018

Oral NAC + sodium chloride 0.9% vs sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.9%

Sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

serious²

none

129/1525

(8.5%)

145/1534

(9.5%)

RR 0.89 (0.71 to 1.12)

1 fewer per 100 (from 3 fewer to 1 more)

MODERATE

1: Castini 2010; Chong 2015; Hafiz 2012; Weisbord 2018
2: 95% confidence interval crosses one end of a defined MID interval

Other outcomes: Oral NAC + sodium chloride 0.9% vs sodium bicarbonate

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.9%	Sodium bicarbonate	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (30 days)
1	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/153 (0.65%)	0/157 (0%)	RR 3.08 (0.13 to 74.98)	-	VERY LOW
Mortality: all-cause mortality (90 days)
1	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	40/1238 (3.2%)	33/1254 (2.6%)	RR 1.23 (0.78 to 1.93)	1 more per 100 (from 1 fewer to 2 more)	LOW
Need for renal replacement therapy: dialysis
1	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	14/1238 (1.1%)	16/1254 (1.3%)	RR 0.89 (0.43 to 1.81)	0 fewer per 100 (from 1 fewer to 1 more)	LOW

1: Chong 2015
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Weisbord 2018

Sensitivity analysis excluding studies with a high risk of bias: Oral NAC + sodium chloride 0.9% vs sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.9%

Sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

serious²

none

119/1372

(8.7%)

126/1385

(9.1%)

RR 0.95 (0.75 to 1.21)

0 fewer per 100 (from 2 fewer to 2 more)

MODERATE

1: Castini 2010; Hafiz 2012; Weisbord 2018
2: 95% confidence interval crosses one end of a defined MID interval

Oral NAC + sodium chloride 0.9% vs oral NAC + sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.9%

Oral NAC + sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

serious²

no serious indirectness

serious³

none

166/2020

(8.2%)

207/2036

(10.2%)

RR 0.81 (0.67 to 0.98)

2 fewer per 100 (from 0 fewer to 3 fewer)

LOW

1: Briguori 2007; Chong 2015; Hafiz 2012; Lee 2011; Maioli 2008; Weisbord 2018
2: i-squared >33.3%
3: 95% confidence interval crosses one end of a defined MID interval

Subgroup analyses (outcome: CI-AKI): Oral NAC + sodium chloride 0.9% vs oral NAC + sodium bicarbonate

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.9%	Oral NAC + sodium bicarbonate	Relative (95% CI)	Absolute	Quality
Subgroup: diabetes
1	randomised trials¹	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	12/59 (20.3%)	8/62 (12.9%)	RR 1.58 (0.69 to 3.58)	7 more per 100 (from 4 fewer to 33 more)	LOW
Subgroup: low volume of contrast agent
1	randomised trials³	no serious risk of bias	no serious inconsistency	no serious indirectness	serious⁴	none	2/137 (1.5%)	8/134 (6%)	RR 0.24 (0.05 to 1.13)	45 fewer per 1000 (from 57 fewer to 8 more)	MODERATE
Subgroup: high volume of contrast agent
2	randomised trials⁵	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	18/135 (13.3%)	16/131 (12.2%)	RR 1.11 (0.59 to 2.09)	1 more per 100 (from 5 fewer to 13 more)	LOW

1: Maioli 2008
2: 95% confidence interval crosses both ends of a defined MID interval
3: Lee 2011 did not provide a definition of low volume of contrast agent
4: 95% confidence interval crosses one end of a defined MID interval
5: Lee 2011 reported high volume of contrast agent ≥140 mL and >5 times body weight (kg) per serum creatinine (ml/dl); Maioli 2008 reported high volume of contrast agent >140ml

Other outcomes: Oral NAC + sodium chloride 0.9% vs oral NAC + sodium bicarbonate

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.9%	Oral NAC + sodium bicarbonate	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (30 days)
2	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/342 (0.29%)	3/349 (0.86%)	RR 0.44 (0.06 to 2.94)	0 fewer per 100 (from 1 fewer to 2 more)	VERY LOW
Mortality - All-cause mortality (30 days–6 months)
3	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	serious⁵	none	45/1679 (2.7%)	36/1700 (2.1%)	RR 1.27 (0.82 to 1.95)	1 more per 100 (from 0 fewer to 2 more)	MODERATE
Need for renal replacement therapy: dialysis
1	randomised trials⁶	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	14/1238 (1.1%)	16/1257 (1.3%)	RR 0.89 (0.44 to 1.81)	0 fewer per 100 (from 1 fewer to 1 more)	LOW

1: Chong 2015; Lee 2011
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Lee 2011 (30 days - 6 months); Maioli 2008 (follow-up time for mortality not reported); Weisbord 2018 (3 months)
5: 95% confidence interval crosses one end of a defined MID interval
6: Weisbord 2018

Sensitivity analysis excluding studies with a high risk of bias: Oral NAC + sodium chloride 0.9% vs oral NAC + sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.9%

Oral NAC + sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

serious²

no serious indirectness

serious³

none

156/1867

(8.4%)

191/1885

(10.1%)

RR 0.83 (0.67 to 1.01)

2 fewer per 100 (from 3 fewer to 0 more)

LOW

1: Briguori 2007; Hafiz 2012; Lee 2011; Maioli 2008; Weisbord 2018
2: i-squared >33.3%
3: 95% confidence interval crosses one end of a defined MID interval

Oral NAC + sodium chloride 0.9% vs IV NAC + sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium chloride 0.9%

IV NAC + sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

very serious³

none

14/102

(13.7%)

13/102

(12.7%)

RR 1.08 (0.53 to 2.18)

1 more per 100 (from 6 fewer to 15 more)

VERY LOW

1: Erturk 2014
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: Oral NAC + sodium chloride 0.9% vs IV NAC + sodium chloride 0.9%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium chloride 0.9%	IV NAC + sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (30 days)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	0/102 (0%)	1/102 (0.98%)	RR 0.33 (0.01 to 8.09)	1 fewer per 100 (from 1 fewer to 7 more)	VERY LOW
Mortality: all-cause mortality (1 year)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	8/102 (7.8%)	12/102 (11.8%)	RR 0.67 (0.28 to 1.56)	4 fewer per 100 (from 8 fewer to 7 more)	VERY LOW
Need for renal replacement therapy: dialysis
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/102 (0.98%)	0/102 (0%)	RR 3 (0.12 to 72.79)	-	VERY LOW

1: Ertuk 2014
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval

Oral NAC + sodium bicarbonate vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium bicarbonate

Sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

serious²

none

134/1337

(10%)

115/1324

(8.7%)

RR 1.15 (0.91 to 1.45)

1 more per 100 (from 1 fewer to 4 more)

MODERATE

1: Hafiz 2012; Weisbord 2018
2: 95% confidence interval crosses one end of a defined MID interval

Other outcomes: Oral NAC + sodium bicarbonate vs sodium chloride 0.9%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium bicarbonate	Sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (90 days)
1	randomised trials¹	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	27/1257 (2.1%)	28/1244 (2.3%)	RR 0.95 (0.57 to 1.61)	0 fewer per 100 (from 1 fewer to 1 more)	LOW
Need for renal replacement therapy: dialysis
1	randomised trials¹	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	16/1257 (1.3%)	15/1244 (1.2%)	RR 1.06 (0.52 to 2.13)	0 more per 100 (from 1 fewer to 1 more)	LOW

1: Weisbord 2018
2: 95% confidence interval crosses both ends of a defined MID interval

Oral NAC + sodium bicarbonate vs sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium bicarbonate

Sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

serious²

none

151/1566

(9.6%)

140/1564

(9%)

RR 1.08 (0.86 to 1.34)

1 more per 100 (from 1 fewer to 3 more)

MODERATE

1: Caglar 2014; Chong 2015; Hafiz 2012; Heng 2008; Weisbord 2018
2: 95% confidence interval crosses one end of a defined MID interval

Other outcomes: Oral NAC + sodium bicarbonate vs sodium bicarbonate

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Oral NAC + sodium bicarbonate	Sodium bicarbonate	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (30 days)
1	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	2/156 (1.3%)	0/157 (0%)	RR 5.03 (0.24 to 103.97)	-	VERY LOW
Mortality: all-cause mortality (90 days)
1	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	27/1257 (2.1%)	33/1254 (2.6%)	RR 0.82 (0.49 to 1.35)	0 fewer per 100 (from 1 fewer to 1 more)	LOW
Need for renal replacement therapy: dialysis
1	randomised trials⁴	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	16/1257 (1.3%)	16/1254 (1.3%)	RR 1 (0.5 to 1.99)	0 fewer per 100 (from 1 fewer to 1 more)	LOW
Adverse events
1	randomised trials⁵	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	1/28 (3.6%)	0/32 (0%)	RR 3.41 (0.14 to 80.59)	-	VERY LOW

1: Chong 2015
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Weisbord 2018
5: Heng 2008 (congestive heart failure)

Sensitivity analysis excluding studies with a high risk of bias: Oral NAC + sodium bicarbonate vs sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Oral NAC + sodium bicarbonate

Sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

serious²

none

134/1387

(9.7%)

119/1383

(8.6%)

RR 1.12 (0.89 to 1.42)

1 more per 100 (from 1 fewer to 4 more)

MODERATE

1: Caglar 2014; Hafiz 2012; Weisbord 2018
2: 95% confidence interval crosses one end of a defined MID interval

IV NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV NAC + sodium chloride 0.45%

Sodium chloride 0.45

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

serious³

no serious indirectness

very serious⁴

none

15/190

(7.9%)

28/194

(14.4%)

RR 0.46 (0.16 to 1.36)

8 fewer per 100 (from 12 fewer to 5 more)

VERY LOW

1: Carbonell 2007; Carbonell 2010; Poletti 2007
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: i-squared >33.3%
4: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: IV NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	IV NAC + sodium chloride 0.45%	Sodium chloride 0.45	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (1 year)
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	6/39 (15.4%)	9/42 (21.4%)	RR 0.72 (0.28 to 1.83)	6 fewer per 100 (from 15 fewer to 18 more)	VERY LOW
Mortality: in-hospital mortality
2	randomised trials⁴	serious²	no serious inconsistency	no serious indirectness	very serious³	none	7/146 (4.8%)	12/151 (7.9%)	RR 0.61 (0.25 to 1.5)	3 fewer per 100 (from 6 fewer to 4 more)	VERY LOW
Need for renal replacement therapy: dialysis
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	very serious³	none	0/39 (0%)	1/42 (2.4%)	RR 0.36 (0.02 to 8.54)	2 fewer per 100 (from 2 fewer to 18 more)	VERY LOW

1: Carbonell 2010 (chronic renal disease arm)
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Carbonell 2007 (normal renal function arm); Carbonell 2010 (chronic renal disease arm)

IV NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV NAC + sodium chloride 0.9%

Sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

serious³

none

129/975

(13.2%)

122/940

(13%)

RR 1.05 (0.84 to 1.32)

1 more per 100 (from 2 fewer to 4 more)

LOW

1: Erturk 2014; Jaffery 2012; Kama 2014; Koc 2012; Kotlyar 2005; Rashid 2004; Traub 2013; Turedi 2016; Webb 2004
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: 95% confidence interval crosses one end of a defined MID interval

Subgroup analyses (outcome: CI-AKI): IV NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	IV NAC + sodium chloride 0.9%	Sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Subgroup: chronic kidney disease
1	randomised trials¹	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	6/57 (10.5%)	6/41 (14.6%)	RR 0.72 (0.25 to 2.07)	4 fewer per 100 (from 11 fewer to 16 more)	LOW
Subgroup: diabetes
1	randomised trials³	serious⁴	no serious inconsistency	no serious indirectness	very serious²	none	2/80 (2.5%)	3/80 (3.8%)	RR 0.67 (0.11 to 3.88)	1 fewer per 100 (from 3 fewer to 11 more)	VERY LOW
Subgroup: older people >75 years
1	randomised trials⁵	serious⁴	no serious inconsistency	no serious indirectness	very serious²	none	0/80 (0%)	6/80 (7.5%)	RR 0.08 (0 to 1.34)	7 fewer per 100 (from 8 fewer to 3 more)	VERY LOW
Subgroup: high volume of contrast agent
2	randomised trials⁶	serious⁴	serious⁷	no serious indirectness	serious²	none	5/98 (5.1%)	10/89 (11.2%)	RR 0.5 (0.08 to 3.18)	56 fewer per 1000 (from 103 fewer to 245 more)	VERY LOW

1: Jaffery 2012
2: 95% confidence interval crosses both ends of a defined MID interval
3: Koc 2012
4: >33.3% of weighted data from studies at moderate or high risk of bias
5: Koc 2012 reported older people ≥70 years
6: Jaffery 2012 reported high volume of contrast agent >300 ml; Koc 2012 reported high volume of contrast agent >100ml
7: I-Squared >33.3%

Other outcomes: IV NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	IV NAC + sodium chloride 0.9%	Sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Mortality: all-cause mortality (up to 8 days)
1	randomised trials¹	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	7/220 (3.2%)	5/227 (2.2%)	RR 1.44 (0.47 to 4.48)	1 more per 100 (from 1 fewer to 8 more)	VERY LOW
Mortality: all-cause mortality (up to 30 days)
3	randomised trials⁴	very serious²	no serious inconsistency	no serious indirectness	very serious³	none	7/528 (1.3%)	10/522 (1.9%)	RR 0.69 (0.27 to 1.81)	1 fewer per 100 (from 1 fewer to 2 more)	VERY LOW
Mortality: all-cause mortality (1 year)
1	randomised trials⁵	serious⁶	no serious inconsistency	no serious indirectness	very serious³	none	12/102 (11.8%)	7/103 (6.8%)	RR 1.73 (0.71 to 4.22)	5 more per 100 (from 2 fewer to 22 more)	VERY LOW
Mortality: in-hospital mortality
2	randomised trials⁷	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious³	none	12/291 (4.1%)	13/279 (4.7%)	RR 0.94 (0.45 to 1.96)	0 fewer per 100 (from 3 fewer to 4 more)	LOW
Need for renal replacement therapy
3	randomised trials⁸	no serious risk of bias	serious⁹	no serious indirectness	very serious³	none	11/223 (4.9%)	17/225 (7.6%)	RR 0.68 (0.34 to 1.36)	2 fewer per 100 (from 5 fewer to 3 more)	VERY LOW
Length of hospital stay in days (Better indicated by lower values)
1	randomised trials¹⁰	no serious risk of bias	no serious inconsistency	no serious indirectness	serious¹¹	none	192	206	-	MD 0.40 lower (0.98 lower to 0.18 higher)	MODERATE

1: Webb 2014
2: >33.3% of weighted data from studies at high risk of bias
3: 95% confidence interval crosses both ends of a defined MID interval
4: Ertuk 2014 (30 days); Jaffrey 2012 (30 days); Webb 2014 (>8 days)
5: Ertuk 2014
6: >33.3% of weighted data from studies at moderate or high risk of bias
7: Jaffrey 2012; Turedi 2016
8: Ertuk 2014 (dialysis); Kama 2014 (renal replacement therapy due to CI-AKI); Turedi 2016 (hemodialysis or peritonael dialysis requirement for severe renal failure)
9: i-squared >33.3%
10: Jaffery 2012
11: 95% confidence interval crosses one end of a defined MID interval

Sensitivity analysis excluding studies with a high risk of bias: IV NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV NAC + sodium chloride 0.9%

Sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

serious³

no serious indirectness

very serious⁴

none

92/781

(11.8%)

88/736

(12%)

RR 1.02 (0.78 to 1.33)

0 more per 100 (from 3 fewer to 4 more)

VERY LOW

1: Erturk 2014; Jaffery 2012; Kama 2014; Koc 2012; Kotlyar 2005; Rashid 2004; Traub 2013; Turedi 2016
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: i-squared >33.3%
4: 95% confidence interval crosses both ends of a defined MID interval

IV NAC + sodium chloride 0.9% vs sodium chloride 0.9% + sodium bicarbonate

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV NAC + sodium chloride 0.9%

Sodium chloride 0.9% + sodium bicarbonate

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

very serious²

none

27/121

(22.3%)

22/121

(18.2%)

RR 1.23 (0.74 to 2.03)

4 more per 100 (from 5 fewer to 19 more)

LOW

1: Kama 2014; Turedi 2016
2: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: IV NAC + sodium chloride 0.9% vs sodium chloride 0.9% + sodium bicarbonate

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	IV NAC + sodium chloride 0.9%	Sodium chloride 0.9% + sodium bicarbonate	Relative (95% CI)	Absolute	Quality
Mortality: in-hospital mortality
1	randomised trials¹	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	11/85 (12.9%)	10/85 (11.8%)	RR 1.1 (0.49 to 2.45)	1 more per 100 (from 6 fewer to 17 more)	LOW
Need for renal replacement therapy
2	randomised trials³	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none	11/121 (9.1%)	11/121 (9.1%)	RR 1 (0.45 to 2.22)	0 fewer per 1000 (from 50 fewer to 111 more)	LOW
2	randomised trials³	no serious risk of bias	no serious inconsistency	no serious indirectness	very serious²	none		8.1%	RR 1 (0.45 to 2.22)	0 fewer per 1000 (from 45 fewer to 99 more)	LOW

1: Turedi 2016
2: 95% confidence interval crosses both ends of a defined MID interval
3: Kama 2014 (renal replacement therapy due to CI-AKI); Turedi 2016 (hemodialysis or peritonael dialysis requirement for severe renal failure)

IV NAC bolus + oral NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV NAC bolus + oral NAC

Control

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

very serious²

very serious³

no serious indirectness

very serious⁴

none

54/341

(15.8%)

62/218

(28.4%)

RR 0.61 (0.21 to 1.83)

11 fewer per 100 (from 22 fewer to 24 more)

VERY LOW

1: Aslanger 2012; Marenzi 2006
2: >33.3% of weighted data from studies at high risk of bias
3: No explanation was provided
4: 95% confidence interval crosses both ends of a defined MID interval

Other outcomes: IV NAC bolus + oral NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	IV NAC bolus + oral NAC + IV sodium chloride 0.9%	IV sodium chloride 0.9%	Relative (95% CI)	Absolute	Quality
Mortality: in-hospital mortality
1	randomised trials¹	serious²	no serious inconsistency	no serious indirectness	no serious imprecision	none	8/233 (3.4%)	13/119 (10.9%)	RR 0.31 (0.13 to 0.74)	8 fewer per 100 (from 3 fewer to 10 fewer)	MODERATE
Need for renal replacement therapy
1	randomised trials³	serious²	no serious inconsistency	no serious indirectness	serious⁴	none	3/189 (1.6%)	6/119 (5%)	RR 0.31 (0.08 to 1.23)	3 fewer per 100 (from 5 fewer to 1 more)	LOW

1: Marenzi 2006
2: >33.3% of weighted data from studies at moderate or high risk of bias
3: Marenzi 2006 (acute renal failure requiring renal replacement therapy)
4: 95% confidence interval crosses one end of a defined MID interval

Sensitivity analysis excluding studies with a high risk of bias: IV NAC bolus + oral NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV NAC bolus + oral NAC + IV sodium chloride 0.9%

IV sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

serious²

no serious inconsistency

no serious indirectness

no serious imprecision

none

27/233

(11.6%)

39/119

(32.8%)

RR 0.35 (0.23 to 0.55)

21 fewer per 100 (from 15 fewer to 25 fewer)

MODERATE

1: Marenzi 2006
2: >33.3% of weighted data from studies at moderate or high risk of bias

IV NAC (bolus) + IV sodium chloride 0.9% vs IV sodium chloride 0.9%

Quality assessment

No of patients

Effect

Quality

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

IV NAC (bolus) + IV sodium chloride 0.9%

IV sodium chloride 0.9%

Relative (95% CI)

Absolute

CI-AKI (as reported by study)

randomised trials¹

no serious risk of bias

no serious inconsistency

no serious indirectness

serious²

none

18/126

(14.3%)

25/123

(20.3%)

RR 0.70 (0.4 to 1.22)

6 fewer per 100 (from 12 fewer to 4 more)

MODERATE

1: Thiele 2010
2: 95% confidence interval crosses one end of a defined MID interval

Network meta-analyses

No. of studies	Study design	Sample size	Effect estimate	Risk of bias	Indirectness	Inconsistency	Imprecision	Quality
CI-AKI
70	RCT	21,825	See appendix G	Serious¹	Not serious	Serious²	No serious	Low

1: >33.3% of studies in the NMA at moderate or high risk of bias
2: DIC for a random-effects model lower than the DIC for a fixed-effects model

Appendix I. Economic evidence study selection

Appendix J. Economic evidence tables

None – no economic evaluations relevant to the review question were found.

Appendix K. Health economic evidence profiles

None – no economic evaluations relevant to the review question were found.

Appendix L. Health economic analysis

Introduction

We did not find any relevant published cost–utility analyses; therefore, we undertook health economic modelling to answer the review question in Table 13. The developers of the previous iteration of the guideline (CG169) created a model to answer this question. After discussion with the committee, we agreed to adapt this existing model as it was deemed to be suitable for decision-making.

Table 13. Research question addressed by economic model

Methods

Model overview

Modelled population(s), intervention(s), comparator(s) and outcome(s)

In the previous iteration of the model, the base-case population was chronic kidney disease (CKD) stage 3–4. The committee was happy that CKD was a good representation of a population ‘at risk’ of CI-AKI to answer the review question. After reviewing the natural history data available to us relating to CKD progression, mortality and the probability of end-stage renal disease and death following CI-AKI, we adapted the original population slightly to also incorporate those patients who had pre-dialysis stage 5 CKD. This was because much of the published data are reported in terms of progression to renal replacement therapy (RRT; a sub-set of stage 5) rather than progression to CKD stage 5 as a whole. A portion of people with stage 5 CKD may be classed as ‘pre-dialysis’; these people are now included in the initial state and can progress to RRT, as can people in stages 3 and 4.

In alignment with the previous model, we used percutaneous coronary intervention (PCI) as a proxy for the probability of repeat scans, as this is a common procedure in which people with CKD are likely to receive iodine based contrast media. The interventions and comparators were determined by the randomised controlled trials (RCTs) included in the clinical review. Quality-adjusted life-years (QALYs) were derived using NICE’s preferred methods. Table 14 summarises the modelled population, interventions, comparators and outcomes.

Table 14. Economic Model PICO

Type of evaluation, time horizon, perspective, discount rate

As per the NICE Reference Case, this evaluation is a cost–utility analysis (reporting health benefits in terms of QALYs), conducted from the perspective of the NHS/PSS, which assesses costs and health benefits using a lifetime horizon, and uses a discount rate of 3.5% per annum for both costs and health benefits.

Model structure

In agreement with the committee, we adapted the model structure from the previous guideline. The model uses a Markov structure with states based on CKD stage or CI-AKI. The committee agreed that a three-month cycle length is appropriate to capture relevant events and changes between states. Figure 5 illustrates the model structure.

Figure 5. Model structure

There are 4 health states within the model (blue boxes): stage 3–5 CKD (pre-dialysis), CI-AKI, permanent RRT and death (see Table 15 for a summary of these health states). At model initiation, all people are in stage 3–5 CKD (pre-dialysis) and start by undergoing a scan, represented by the green arrow in Figure 5. As a result of this scan they can either develop CI-AKI and move to the corresponding state, or they are assumed to have no complications and return to the ‘Stage 3–5 CKD’ state. The risk of CI-AKI following a scan (indicted by red arrow) is a key model parameter; it is obtained from the network meta-analysis (NMA) undertaken as part of the clinical review and represents the relative effectiveness of each of the interventions in terms of risk of CI-AKI. People who develop CI-AKI following a scan can either return to the ‘Stage 3–5 CKD’ state (assuming complete resolution of CI-AKI), require permanent RRT as a result of their CI-AKI, or die within the cycle (the CI-AKI increases this risk). Those people who return to the ‘Stage 3–5 CKD’ state following a scan or an episode of CI-AKI may need repeat scans, which have the same potential consequences as the first scan, or they may progress to permanent RRT.

People who are in the permanent RRT state can either be on dialysis or receive a kidney transplant (note that the possibility of transplantation was not included in the previous iteration of the model). Simulated people stay in this state until they die and are assumed not to have repeat contrast-enhanced scans. The committee advised that some people on dialysis may undergo scans using contrast; however, this would be avoided where possible and therefore this simplifying assumption was acceptable for decision making purposes. CI-AKI in renal transplant patients is not within the scope of the update; therefore, we assumed that anyone who has a transplant will not undergo repeat contrast-enhanced scans.

Table 15. Modelled health states

Each of the health states (apart from death) is associated with costs and QALYs, which accumulate over the model horizon. A half-cycle correction is applied to account for the fact that people may transition from one state to another at any point throughout a cycle, rather than only at the start.

Model parameterisation

Identifying sources of parameters

With the exception of treatment effects, which were comprehensively updated (see below), we used the parameters from the previous iteration of the model unless we could find anything more appropriate or recent from informal searches. These informal searches aimed to satisfy the principle of ‘saturation’ (that is, to ‘identify the breadth of information needs relevant to a model and sufficient information such that further efforts to identify more information would add nothing to the analysis’ [Kaltenthaler et al., 2011]). We conducted searches in a variety of general databases, including Medline (via PubMed) and GoogleScholar. Where we could not identify suitable evidence from informal searches or it was not present/appropriate in the existing model, sources for parameters were also sought from the guideline committee directly. Any key parameters that were different to the previous iteration of the model were validated by the committee.

Selecting parameters

Our overriding selection criteria were as follows:

The selected studies should report outcomes that correspond as closely as possible to the health states and events simulated in the model.
The selected studies should report a population that closely matches the UK population (ideally, they should be drawn from the UK population).
All other things being equal, we prefer more powerful studies (based on sample size and/or number of events).
Where there was no reason to discriminate between multiple possible sources for a given parameter, we gave consideration to quantitative synthesis (meta-analysis), to provide a single summary estimate.

Parameters

Key calculations and parameters are summarised here. Please see the full table of parameters (Table 33) for a complete summary of all parameters used in the model, including their distributions and sources.

Cohort parameters and natural history

Initial cohort settings

The base-case cohort has stage 3–5 CKD (pre-dialysis) to represent a population who are ‘at risk’ of CI-AKI. The cohort is 50% male and is aged 70 years in line with the previous iteration of the model.

Natural progression to permanent RRT

The previous iteration of the model used a Norwegian study to obtain the probability of progressing from stages 3–4 CKD to stage 5 CKD (Eriksen & Ingebretsen, 2006). This study has two main limitations: firstly, it is from a non-UK population and, secondly, it only includes people with stage 3 CKD at baseline (therefore, it may underestimate rates of progression). We were able to find a UK study that allowed us to obtain the probability of progressing from stages 3–5 CKD to RRT (Marks et al., 2012). This large, UK study included 3,414 patients with CKD stages 3–5 who were not on RRT at baseline. The study reports rates of RRT initiation after 6 years of follow-up, allowing us to calculate the 3-month probability of transitioning to the permanent RRT state (Table 16). In the 2 oldest age categories (age 85–94 years, and 95–104 years) there were no events within the study; therefore, we used linear extrapolation based on observed rates (on the log scale) in the younger age groups to obtain predicted rates in the older two age groups. At base-case values, the regression equation is: ln(rate) = −0.91 – 0.05age.

Table 16. Progression from stages 3–5 CKD to RRT (Marks et al., 2012)

Mortality in stages 3–5 CKD

We used the Norwegian study by Eriksen and Ingebretsen (2006) for mortality in stages 3–5 CKD, as we were unable to find evidence from the UK that reported data in a form suitable for use within the model. All included participants within the study had stage 3 CKD at baseline so did not exactly match our modelled population of stages 3–5 CKD, and therefore mortality rates may be underestimated. The study reports data as standardised mortality ratios (SMRs) relative to the general population (Table 17). Our model applies these SMRs to National Life Table data for England and Wales (Office for National Statistics, 2018) to obtain age-specific probabilities of death in stages 3–5 CKD.

Table 17. Standardised mortality ratios versus the general population in stages 3–5 CKD (Eriksen & Ingebretsen, 2006)

Mortality in RRT

The previous model used a French study to obtain standardised mortality ratios versus the general population in people on dialysis (Villar et al., 2007). Following advice from the committee, we updated the model to include people who have received kidney transplants within the permanent RRT state. Therefore, we wanted mortality data to reflect this. The UK Renal Registry (UK Renal Registry, 2018) provided us with the relative risk of death in people receiving RRT compared with the general population in the 2016 registry cohort (Table 18), which we applied to UK Life Table data (ONS) to obtain the probability of death in the permanent RRT state by age. The Renal Registry reports a relative risk of 1.5 in the 85+ years age group; however, when applied to life table data, this resulted in people in the RRT state having a lower probability of death than those in the CKD 3–5 state, which lacked face-validity. We note that Renal Registry data for other years do not feature a drop-off in risk of the same magnitude, which further suggests that this finding is artefactual. We therefore made an assumption that the relative risk in the 85+ years group was the same as the 80–84 years group.

Table 18. Relative risk of death compared with the UK general population in prevalent renal replacement therapy patients (UK Renal Registry, 2018)

Baseline risk of CI-AKI

We explored a number of different sources for the baseline risk of CI-AKI. The base-case rate was obtained from an Italian study of 502 people with CKD (Maioli et al., 2008). We used this as the base case because it was European, the population was appropriate for our modelled cohort (CKD) and it was very similar to the study we used to obtain mortality data (same centre, investigators and interventions). In this study, 29 out of a total of 252 people (11.5%) developed CI-AKI in the sodium chloride 0.9% + NAC arm.

We explored the impact of using estimates from other individual studies in sensitivity analysis. The only UK study that was included in the clinical review (Rashid et al., 2004) included a small subset of people (n=21) who had raised serum creatinine. Of these, 3 (14.3%) experienced CI-AKI in the sodium chloride 0.9% arm. We also explored the 3 different rates that were used in the previous iteration of the model: a low rate of 2.2% (Mueller et al., 2002), a medium rate of 19.2% (Dangas et al., 2005) and a high rate of 30.0% (Mehran et al., 2004).

We noted that baseline rates of CI-AKI differed substantially between the studies; for example, in the sodium chloride 0.9% study arms the rate varied from 2.2% (Mueller et al., 2002) to 36.7% (Sadineni et al., 2017). In an attempt to account for this, we conducted a supplementary analysis in which we synthesised the rates across all sodium chloride 0.9% study arms. This interventional arm was chosen for the baseline synthesis because it had the greatest amount of data available. The rate of CI-AKI from this pooled analysis was 13.1%, which was in reasonably close alignment with the rates from the Maioli et al., (2008) and Rashid et al., (2004) studies discussed above. To better understand the variation in rates of CI-AKI between trials, we stratified the studies with sodium chloride 0.9% arms into elective and emergency and conducted separate pooled analyses, as we would expect emergency patients to have higher overall morbidity and therefore be at an overall higher risk of CI-AKI. The pooled rates of CI-AKI were 10.8% in elective settings and 19.6% in emergency settings. Funnel plots showing the rates of CI-AKI in 0.9% sodium chloride arms against the number of participants in the study are shown for the base case (Figure 6), the elective population (Figure 7) and the emergency population (Figure 8). These funnel plots show the risk of CI-AKI against the number of participants in each study who received sodium chloride 0.9%. The plots display the increasing precision in the estimate as the study size increases. They show that most of the observed variability in event-rates is to be expected, given the imprecision inherent in small sample sizes: relatively few of the estimates lie outside the expected uncertainty intervals.

Figure 6. Funnel plot, rates of CI-AKI in 0.9% sodium chloride arms, base case

Figure 7. Funnel plot, rates of CI-AKI in 0.9% sodium chloride arms, elective studies

Figure 8. Funnel plot, rates of CI-AKI in 0.9% sodium chloride arms, emergency studies

Table 19 summarises the different baseline rates of CI-AKI; we explore how varying the baseline rate of CI-AKI affects results in sensitivity analyses.

Table 19. Baseline risk of CI-AKI

Risk of permanent RRT following CI-AKI

Following a case of CI-AKI there will be a small subset of people whose renal function will not recover and who will require permanent RRT. A study by James et al., (2011) reported the incidence of end-stage renal disease (ESRD; as defined by the requirement for dialysis or transplantation) following cases of CI-AKI, although this was not stratified according to CKD status. However, the study reports rates of CI-AKI in people with and without CKD, and the rates of ESRD per 100 person-years in people with and without CKD (see the full list of parameters in Table 33). From this, we estimate that the probability of transitioning to the permanent RRT state following CI-AKI is 4.08%. This is in reasonably close alignment with the figure used in the previous version of the model (3.28%; James et al., 2010); however we did not use this value as we were unable to replicate the calculations undertaken by the previous authors.

All-cause mortality following CI-AKI

We obtained mortality data from a European study of 1,490 people with CKD undergoing coronary angiography (Maioli et al., 2012; also see ‘Baseline risk of CI-AKI’ above). In total, 180 people experienced CI-AKI and 1,310 did not. In the 3 months following the coronary angiography, 13 people (7.22%) died in the CI-AKI group, while 18 people (1.37%) died in the no CI-AKI group. We used these data to calculate both the OR (5.59) and risk difference (5.85%) for mortality in the CI-AKI versus no CI-AKI group (Table 20). In the base case, we applied the OR to the background probability of death in people with CKD stage 3–5, thereby assuming there is a relative hazard of death associated with CI-AKI compared with no CI-AKI. In a sensitivity analysis, we applied the risk difference to the probability of death in CKD stage 3–5, assuming that there is an absolute excess hazard of death.

As well as the study by Maioli et al., (2012), we explored two additional sources of mortality data. A retrospective cohort study by Hoste et al., (2011) evaluated the epidemiology of CI-AKI in intensive care patients. Not all patients had CKD, so the study is not directly applicable to our population of interest; however, it does allow us to explore mortality in a critically ill emergency population.

Finally, we extracted mortality data from the study by James et al., (2011) that was used to obtain the risk of permanent RRT following CI-AKI. Although the data were not reported directly, we were able to use the incidence of mortality in all patients (CKD and non-CKD), the proportion of people with CKD in the cohort, and the rate of mortality (per 100 person-years) in people with CKD to estimate the values needed (see the full list of parameters in Table 33 for values). A summary of mortality ORs and risk differences from all sources used within the model is presented in Table 20.

Table 20. Mortality following CI-AKI

Probability of repeat scans

The previous version of the model used the probability of a repeat PCI as a surrogate for a repeat scan. The committee was happy with this approach, so we replicated it in our analysis. The annual probability of a repeat PCI used in the previous version of the model was 11.4%, taken from a study by Serruys et al., (2009); however, this study population was not specific to CKD. We identified a directly relevant Canadian study by Chan et al., (2015), investigating rates of repeat PCI in people with CKD who originally received PCI for revascularisation for multivessel disease. Of 893 patients who underwent the original procedure, 131 required repeat revascularisation over an average of 1.76 years of follow-up. From this, we calculated the per-cycle rate of repeat PCI to be 2.06%.

Treatment effects

An NMA was undertaken to combine direct and indirect evidence on the effectiveness of each intervention in preventing CI-AKI (see Appendix B – Methods and Appendix G – Network meta-analysis results). We obtained relative treatment effects from the NMA as log ORs (lnORs) versus no intravenous hydration. We re-expressed these lnORs as relative to oral NAC plus sodium chloride 0.9% by subtracting each of them from the lnOR for that strategy. We were then able to apply these lnORs to the base-case baseline risk of CI-AKI (11.5%; see ‘Baseline risk of CI-AKI’ above) to obtain the absolute odds of CI-AKI for each intervention. We then converted these odds to probabilities (Table 21). In sensitivity analyses, we explored numerous sources for the baseline risk of CI-AKI, each of which included different interventions (see above). We therefore altered the baseline intervention according to which was used in the source data and applied the relative treatment effects to the appropriate baseline risk.

Table 21. Treatment effects

Resource use and costs

With some exceptions (described in detail below), we replicated the approach to resource use and costing used in the previous version of the guideline. We split the costs into 2 separate categories: fluid strategies and health states. As a general principle, if we were not able to obtain an error estimate for costs, we assumed the costs were fixed rather than assuming a standard error of 50% of the mean, as was done in the previous iteration of the model.

Fluid strategies: unit costs

The total cost of the fluid strategies incorporates unit costs of the interventions themselves, plus any associated healthcare resource use required for their administration. For intravenous (IV) fluids (sodium chloride 0.9% and 0.45%, and sodium bicarbonate), the previous developers obtained the unit costs either through personal communication from the commercial medicines unit (CMU) or from NHS list prices. We searched a number of sources (the British National Formulary [BNF], the CMU electronic market information tool [eMIT], the NHS Drug Tariff) and contacted the CMU to request costs with national discounts applied. We also inflated the costs used in the previous version of the model as an additional option. We then asked the committee to consider all the available cost estimates to assess which best reflected their experience. Based on committee advice that other costs appeared low, and given that costs were not always listed in the NHS Drug Tariff, we used NHS indicative prices from the BNF as the base case and explored costs provided by the CMU in a sensitivity analysis (Table 22).

Other intervention components include NAC (oral capsules, IV bolus and solution for infusion), sodium bicarbonate capsules, oral sodium citrate and oral fluids. We assumed that there was no cost associated with oral fluids, as this involves asking people to drink a volume of water before their contrast procedure. For the other interventions, we obtained prices from CMU where possible (either personal communication or the eMIT), or alternatively from the NHS Drug Tariff if no price was available from the CMU. We were not able to find a cost for an IV bolus of NAC; therefore, we assumed the same cost as the infusion (Table 22).

Table 22. Intervention unit costs

Fluid strategies: total costs

We used the same infusion strategy volumes and doses as the previous analysis. For interventions that were not previously analysed, we looked at the included trials from the clinical review to determine the appropriate volumes and doses. We varied our analysis slightly compared with the previous version in terms of base case assumptions surrounding hospital admission requirements for IV regimens. The guideline committee for CG169 advised the developers of the previous iteration of the model that regimens involving sodium chloride 0.9% and sodium bicarbonate (apart from the combination of the 2) could be delivered in under 8 hours and therefore did not require overnight hospital admission. However, we were advised by the committee that in most cases admission is required for all IV fluids, particularly in centres that do not have day units or for scans that occur early or late in the day. Therefore, in our base case all IV regimens require admission and therefore attract the cost of an excess bed day. We explored the assumptions from the previous model in a sensitivity analysis. In an additional sensitivity analysis, we assumed that none of the interventions required the cost of an excess bed day. This is to represent an inpatient population who would not encounter any excess bed day costs as a result of contrast administration. Table 23 summarises the components and costs of all intervention strategies.

Table 23. Intervention regimen components and total costs

We used the 2017–18 NHS Schedule of Reference Costs (NHS Improvement, 2018) to estimate the cost of an excess bed day. The previous developers assumed this was an excess bed day for currency code EA36A (Catheter 19 years and over); however, this code was not present in the most recent reference costs so we instead used a pooled average of elective and non-elective excess bed days for cardiac catheterisation (currency codes EY42A–EY43F; pooled average £378.77).

Health states: costing overview

We took a similar approach to costing the health states as the previous model, again with some exceptions. As a general rule, any NHS reference costs were updated to 2017–18 values rather than 2010–11. We also obtained costs and resource data from the Personal Social Services Research Unit (Curtis & Burns, 2018), the British National Formulary (Joint Formulary Committee, 2019), the Commercial Medicines Unit (2019), the NHS Drug Tariff (NHS Business Services Authority, 2019a) and the NICE guideline on RRT and conservative management (NG107; NICE, 2018).

Health states: cost of CI-AKI

The total cost of a cycle in the CI-AKI state consisted of treatment for the AKI, temporary dialysis for a small proportion of people who require it, and permanent RRT for those who progress to RRT following their CI-AKI. We obtained the cost of the AKI treatment and dialysis from NHS reference costs (2017–18). We found 3 studies that report the number of people requiring temporary dialysis following CI-AKI (Kama et al., 2014; Briguori et al., 2002; Briguori et al., 2007); we pooled numbers from each study to calculate the odds and probability of requiring temporary dialysis (4 out of 16 people in Kama et al., 2014; 1 out of 16 people in Briguori et al., 2002; 2 out of 13 people in Briguori et al., 2007). From this, we obtained a probability of 17.53%. We consulted the committee regarding the number of dialysis sessions usually received by people who require temporary dialysis following CI-AKI. We were advised that people usually have 1–3 sessions, and that these sessions are almost always haemodialysis rather than peritoneal dialysis. The cost of CI-AKI is summarised in Table 24. A more detailed summary of the permanent RRT costs is reported in the ‘Health states: cost of RRT’ section.

Table 24. CI-AKI costs and resource use

Health states: cost of CKD stages 3–5 (pre-dialysis)

Costs associated with the CKD stages 3–5 (pre-dialysis) state include nephrology appointments, estimated glomerular filtration rate (eGFR) measurements (consisting of one biochemistry and one phlebotomy test), treatment for anaemia (with epoetin), diuretics, and home and telephone consultations with a nurse (Table 25). We took assumptions surrounding the resource use (e.g. the number of appointments), as well as epoetin and furosemide requirements, from the previous version of the guideline (NICE, 2013a). In terms of the proportion of people in each CKD stage, the previous committee advised that 70% of people were in stage 3, 25% were in stage 4 and 5% were in stage 5 (NICE, 2013a). Of those in stage 5, 39% were assumed to not be on RRT (Hussain et al., 2013). As people on RRT are not included in this health state, the remaining proportions are 72.2% in stage 3, 25.8% in stage 4, and 2.0% in stage 5 (pre-dialysis). We grouped stages 3 and 4 together for costing purposes, while we assume that stage 5 requires greater resource use (e.g. more nephrology appointments per cycle) and therefore attracts a greater cost.

Table 25. CKD stages 3–5 costs and resource use

Health states: cost of RRT

Our approach to costing RRT is heavily based on the RRT guideline (NG107; NICE 2018). Costs within the RRT state consist of appointments, eGFR tests, treatment for anaemia (with epoetin), and costs of the RRT split into dialysis and transplantation. We assume that all patients attract the costs of appointments, eGFR tests and treatment for anaemia, while a proportion of people will have dialysis for the remainder of their lives (either haemodialysis or peritoneal dialysis), and the remainder will receive a transplant. We obtained the overall probability of being waitlisted from the NHS Organ Donation and Transplantation Activity Report (NHS Blood and Transplant, 2018), which we converted to the odds of being waitlisted. We obtained the odds ratios for being wait-listed according to age from the UK Renal Registry (2018). This only went up to age 64, so we used linear extrapolation to calculate the odds ratios for higher age brackets. We used the overall odds of being wait listed and the odds ratios according to age to obtain the probability of being wait-listed for each age group. We obtained the proportions of people on haemodialysis versus peritoneal dialysis from the renal registry (UK Renal Registry, 2018).

Dialysis costs involve an initial access procedure followed by the ongoing costs of dialysis sessions, plus an assumption that 15% of the total dialysis costs are for travel, access maintenance, and other associated costs (NICE, 2018). Transplantation costs include the cost of the transplant itself, plus ongoing immunosuppression. The approach to costing the transplant procedure was adapted from the NICE guideline for renal replacement therapy and conservative management (NICE, 2018), apart from the cost of associated appointments which were already accounted for via the cost of nephrology consultations. The approach to costing immunosuppressant therapy was adapted from the guideline for the management of hyperphosphataemia in chronic kidney disease (NICE, 2013b). As part of this, we assume a one-off induction with basiliximab costing £2,162, followed by ongoing maintenance therapy with azathioprine plus either tacrolimus or ciclosporin costing an average of £1,643 per cycle. See the full parameter table for unit costs of immunosuppressant therapy (Table 33).

Given that we were adapting the model structure from the previous iteration, which did not include transplantation as part of the RRT costs, it was necessary to add or subtract some costs up-front (within the first cycle after entering the RRT state) to ensure the correct proportion of people encountered the costs of transplantation and dialysis at the correct timepoints. Our approach to transplantation costs is summarised below:

Cycle 1
- Apply a one-off cost of transplantation to the proportion of people who are likely to receive a transplant over their lifetime (54%; UK Renal Registry, 2018), discounted at the average transplant waiting list time.
- Apply the costs of dialysis for the time period up to transplantation, with the assumption that the people who do eventually receive a transplant receive dialysis while they are on the waiting list. We discount these dialysis costs continuously up to the point of transplantation.
- People who have a transplant will require immunosuppressant drugs. We subtract the cost of immunosuppressants from the point of RRT initiation to the time of transplant as part of the cycle 1 costs. This is because people who eventually receive a transplant will start to accrue immunosuppressant costs as soon as they enter cycle 2. Unless we subtract some immunosuppressant costs up-front, people will accrue immunosuppressant costs for the time period before they have the transplant and the costs will be overestimated.
- Similarly, we have added the costs of immunosuppressants for one cycle within cycle 1 so that immunosuppressants for the cycle directly following the time of transplant are incorporated.
Cycle 2 onwards
- Apply the costs of ongoing immunosuppressants to the proportion of people who receive a transplant (54%).

Table 26 provides a full summary of RRT costs split according to cycle 1 or cycle 2 onwards.

Table 26. RRT costs and resource use

Quality of life

We were able to find utility values that were more appropriate for our modelled population compared with those used in the previous model. We sourced utility values for CKD stages 3, 4 and 5 (pre-dialysis) from a recent UK study by Jesky et al., (2016), while we obtained the RRT values from a study by Liem et al., (2008), which is in alignment with the NICE clinical guideline on renal replacement therapy and conservative management (NICE, 2018). We did not vary the proportions of people in each of the RRT states by age (as we did for the costs); the proportions were obtained from the UK Renal Registry (2018); 46% of people are on dialysis compared with transplant. Of those who are on dialysis, 87% are receiving haemodialysis and the remaining 13% are receiving peritoneal dialysis.

We were unable to find an appropriate study that reported utility values for CI-AKI in a UK population with CKD. We instead used a Finnish study that reported quality of life measured using the EQ-5D in critically ill people with all types of AKI (Nisula et al., 2013). The study reports utility values at 6 months after admission to intensive care in study participants with AKI and in age- and sex-matched controls. From this, we were able to calculate the relative utility decrement associated with an episode of AKI and apply it to population utility norms (Kind et al., 1999).

Table 27. Utility values

Sensitivity analyses

In order to explore uncertainty in model results, we conducted both deterministic and probabilistic sensitivity analyses.

Deterministic sensitivity analysis

Deterministic analyses either use alternative point estimates for model parameters or test different structural assumptions, in order to investigate the impact on results. The parameters of interest for deterministic sensitivity analysis in the current analysis included:

Baseline risk of CI-AKI
All-cause mortality following CI-AKI
Probability of repeat scans
Assumptions surrounding excess bed day costs

Further to this, we conducted a one-way sensitivity in which all parameters were varied between plausible bounds to determine which have the potential to affect cost-effectiveness results. Finally, we conducted scenario analyses in which we vary 2 or more parameters concurrently. This involved repeating the sensitivity analyses listed above while altering other parameters to represent emergency and elective populations:

Emergency scenario: risk of CI-AKI taken from emergency studies and no excess bed day costs (represents a high-risk inpatient population)
Elective scenario: risk of CI-AKI taken from elective studies, base-case excess bed day cost assumptions (represents a low-risk outpatient population)

Probabilistic sensitivity analyses

We configured the model to perform probabilistic sensitivity analysis to quantify uncertainty in the true values of input parameters. We assigned probability distributions reflecting uncertainty surrounding point estimates to model input parameters. These were defined by standard error/confidence intervals and type of parameter. We sourced distribution parameters from the study in which the value was obtained, where possible, or estimated them based on the usual properties of data of that type. The model draws a random value from each of these distributions for 1,000 iterations and, for each of these iterations, records costs and QALYs for each strategy. This process allows uncertainty around model results to be characterised in terms of the proportion of iterations in which each comparator provides the optimal balance of costs and QALYs at a particular threshold. We can then construct cost-effectiveness acceptability curves (CEACs) to represent these results visually.

The particular distribution assigned to each type of model parameter reflects the nature of the data. As a rule, we use beta distributions to parameterise probabilities, to reflect the fact that these values must lie between 0 and 1. Although the majority of costs within the current model were fixed, some are given a gamma distribution, as these values are bound at 0 but theoretically have no upper limit. We assign a lognormal distribution to relative risks, ORs and hazard ratios, in order to reflect the fact that these parameters are asymmetrically distributed (i.e. values between 0 and 1 favour one comparator, whereas values between 1 and infinity favour the other). As with probabilities, we assign utilities a beta distribution, as they are bounded at 1. For the treatment effects drawn from the NMA, we directly sampled from the WinBUGS CODA output (the posterior estimates of log-odds ratios) to preserve correlation between treatment effects for different interventions.

Original cost–utility model – results

Clinical outcomes

In terms of clinical outcomes, we investigated life-years, rates of CI-AKI, deaths from CI-AKI, progression to ESRD following CI-AKI and numbers of people progressing to ESRD overall for each intervention (Table 28). Results for all clinical outcomes reflect the NMA outputs, with the most effective intervention (sodium bicarbonate + oral fluids) resulting in the greatest number of life-years, and lowest rates of CI-AKI, deaths, ESRD and lifetime RRT. The trends from the NMA continue across all results.

Table 28. Results – clinical outcomes

We explored the rates of CI-AKI following a first scan depending on the initial level of risk (base case [all], elective only or emergency only; Figure 9). The lowest rates were in the elective population, while the highest rates were in the emergency population. This is a direct reflection of the different estimates used for the baseline risk of CI-AKI.

Figure 9. Initial rates of CI-AKI according to baseline risk

Base-case cost–utility results

In the base-case deterministic results, sodium bicarbonate (oral) + oral fluids dominates all other interventions. It has an overall cost of £20,972 and results in 6.395 QALYs. All other interventions are found to be more expensive and less effective. Again, the results directly reflect the NMA outputs, indicating that the risk of CI-AKI is the key parameter underpinning the model.

Table 29. Base-case deterministic cost–utility results

These results can be visualised on the cost-effectiveness plane (Figure 10). The intervention with the lowest cost is placed at the origin (sodium bicarbonate [oral] + oral fluids). All other interventions are located within the north west quadrant in comparison; therefore, all are dominated.

Figure 10. Base-case deterministic cost–utility plane

Sensitivity analysis

Probabilistic sensitivity analysis

The CEAC (Figure 11) shows that, at all values of a QALY, sodium bicarbonate (oral) + oral fluids has the highest probability of being cost effective. As indicated by the cost-effectiveness acceptability frontier (the bold line), the same strategy has the highest expected net benefit at all QALY values.

Figure 11. Cost-effectiveness acceptability curve

One-way sensitivity analysis

Figure 12 presents the results of the one-way sensitivity analysis for sodium bicarbonate (oral) + oral fluids versus oral fluids alone (i.e. the most cost-effective intervention compared with the second most cost-effective). The parameter that has the greatest effect on cost-effectiveness results is the relative effect of sodium bicarbonate (oral) + oral fluids versus no intervention. The 95% credible interval for this parameter was wide, given the high degree of uncertainty surrounding the point estimate; therefore, when varied between the limits of the interval, this parameter has the potential to result in a negative incremental net monetary benefit. No other parameters, when varied, have the possibility of changing the cost-effectiveness conclusion.

Figure 12. One-way sensitivity analysis – sodium bicarbonate (oral) + oral fluids versus oral fluids alone

Other sensitivity analyses

We varied other parameters as described in the methods section. These include:

Using mortality data from a non-CKD population who are critically ill (Hoste et al., 2011) and applying an absolute (rather than relative) risk of mortality. This affected the absolute numbers of costs and QALYs, but did not affect the conclusions of the incremental analysis.
Varying the probability of repeat scans, which did not have a notable impact on results.

Scenario analysis

Without sodium bicarbonate (oral) + oral fluids

Although sodium bicarbonate (oral) + oral fluids was associated with the most positive point-estimate in the NMA and base-case cost-effectiveness results, the committee was not convinced that the evidence is sufficiently robust for it to be recommended. In the NMA, the credible interval surrounding the point estimate for sodium bicarbonate (oral) with oral fluids was very wide, and there was only a single trial arm (comprising 21 participants) contributing to the evidence base (see The committee’s discussion of the evidence). We therefore presented results with this intervention excluded.

When we remove sodium bicarbonate + oral fluids from the decision-making space, oral fluids becomes the cheapest intervention. In an incremental analysis, the only intervention that is not dominated by oral fluids is sodium chloride 0.9% (IV) + sodium bicarbonate (IV), with a very high ICER of £510,922 per QALY gained.

Table 30. Deterministic cost-effectiveness results, without sodium bicarbonate + oral fluids

Figure 13 presents the cost-effectiveness plane for these results. The steep gradient of the red line between intervention 10 (oral fluids) and intervention 15 (sodium chloride 0.9% [IV] + sodium bicarbonate [IV]) represents the ICER of £510,922 per QALY gained. All other interventions remain dominated.

Figure 13. Cost-effectiveness plane, without sodium bicarbonate + oral fluids

Figure 14 presents the CEAC without sodium bicarbonate + oral fluids. It is evident that oral fluids have the highest probability of being cost-effective and the highest expected net benefit across all values of a QALY. The probability that any of the individual intravenous regimens is best is spread thinly among several possible strategies; however, it can be seen to rise somewhat as increasing value is placed on QALY gains.

Figure 14. Cost-effectiveness acceptability curve, without sodium bicarbonate + oral fluids

We conducted an additional PSA in which we group the interventions according to whether they are oral fluids, contain sodium chloride 0.9% and/or sodium bicarbonate, or are something else (e.g. sodium chloride 0.45%, sodium citrate, NAC alone or no treatment). Figure 15 shows that oral fluids have the highest probability of being cost effective when a QALY is valued at less than approximately £25,000, above which regimens containing IV sodium chloride 0.9% and/or sodium bicarbonate are most likely to be cost effective.

Figure 15. Cost-effectiveness acceptability curve, without sodium bicarbonate + oral fluids, grouped regimens

We also undertook an OSA for this scenario, in which we compared sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids (Figure 16). Again, the only parameters that have the potential to alter the cost-effectiveness conclusion are the relative treatment effects; the incremental net monetary benefit becomes positive (that is, the intravenous regimen becomes cost effective) when oral fluids are assumed to be less effective, and when sodium chloride 0.9% (IV) + sodium bicarbonate (IV) is assumed to be more effective.

Figure 16. One-way sensitivity analysis without sodium bicarbonate + oral fluids – sodium chloride 0.9% (IV) + sodium bicarbonate (IV)

To further explore the effects of the baseline risk of CI-AKI on the ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids, we varied the baseline risk between extreme values (1% and 50%; base case 11.5%) in a threshold analysis (Figure 17). No value of the baseline risk of CI-AKI leads to a positive incremental net monetary benefit (INMB); therefore, when all other parameters are evaluated at their base-case value, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) is unlikely to be cost-effective compared with oral fluids at any plausible baseline risk of CI-AKI if a QALY is valued at £20,000.

Figure 17. Threshold analysis – baseline risk of CI-AKI, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids

In an additional threshold analysis, we vary the mortality OR for CI-AKI versus no CI-AKI between extreme values (1 and 10; base case 5.59) to determine the effect on the ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids (Figure 18). The INMB remains negative across all values of the OR, indicating that no plausible value of this parameter leads to sodium chloride 0.9% (IV) + sodium bicarbonate (IV) becoming cost effective if QALYs are valued at £20,000 each.

Figure 18. Threshold analysis – mortality odds ratio for CI-AKI vs no CI-AKI, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids

To determine whether the risk of ESRD following an episode of CI-AKI has the potential to meaningfully affect the ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids, we varied this parameter between extreme values (0% and 10%; base case 4.1%). No value of this parameter leads to a positive INMB (Figure 19).

Figure 19. Results – threshold analysis, probability of ESRD following CI-AKI, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids

Emergency setting

We undertook a scenario analysis for the emergency population (see Methods: Deterministic sensitivity analysis), in which we use the baseline risk of CI-AKI obtained from a synthesis of trials from the emergency setting, and assume that everyone is already an inpatient, so no excess bed day costs are applied for any intervention. We conducted this analysis without sodium bicarbonate + oral fluids, as the committee did not deem this intervention to be an option for recommendation due to the wide credible intervals surrounding the effect point estimate and the small evidence base upon which the the point estimate is based; excluding it from the analysis allowed the committee to better interpret results for the remaining interventions.

Table 31 presents deterministic cost–utility results for the emergency setting. Oral fluids remain the cheapest option; however, in this emergency scenario, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) now has an ICER of £16,112 compared with oral fluids. The differences between these strategies are very small: a QALY gain of 0.0018 is equivalent to around an extra two-thirds of a day in perfect health over an average patient’s lifetime. All other interventions remain dominated. Note that when the 2 emergency assumptions are applied individually, the ICER remains greater than £20,000 per QALY (£26,410 per QALY when the inpatient bed day assumption is applied, and £366,769 per QALY when the emergency baseline risk of CI-AKI is applied).

Table 31. Results – deterministic cost-effectiveness results, emergency setting, without sodium bicarbonate + oral fluids

Figure 20 presents the cost-effectiveness plane for these results. The points representing oral fluids (number 10) and sodium chloride 0.9% (IV) + sodium bicarbonate (IV) are very close together, which is indicative of the small incremental cost and QALY differences between these 2 interventions.

Figure 20. Results – cost-effectiveness plane, emergency setting, without sodium bicarbonate + oral fluids

Figure 21 shows the CEAC. It is evident that, regardless of the value ascribed to QALYs, none of the individual modelled strategies can be identified as optimal with any degree of confidence. Oral fluids appear to have the highest probability of being cost effective, but sodium chloride 0.9% (IV) + sodium bicarbonate (IV) has the greatest expected net benefit if a QALY is valued at £20,000 (Table 31). Oral fluids have the greatest expected net benefit at low QALY values (below approximately £4,000).

Figure 21. Results – CEAC, emergency setting, without sodium bicarbonate + oral fluids

The CEAC for grouped interventions in the emergency setting shows that at all QALY values, regimens containing sodium chloride 0.9% and/or sodium bicarbonate have the highest probability of being cost effective (Figure 22).

Figure 22. Cost-effectiveness acceptability curve, without sodium bicarbonate + oral fluids, grouped regimens – emergency

We undertook an OSA for this scenario (Figure 23), in which we compare sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids. The uncertainty surrounding the cost-effectiveness of sodium chloride 0.9% (IV) + sodium bicarbonate (IV) in this scenario is evidenced by the base case (red dotted line) overlapping with the line indicating an incremental net monetary benefit of £0. Changes to the relative treatment effects (top 2 parameters) have a large effect on results; however, even those parameters that make very little difference to results in absolute terms would still have the potential to change the cost-effectiveness conclusion if a decision-maker were to adopt a rigid threshold of £20,000/QALY.

Figure 23. Results – OSA, emergency setting, without sodium bicarbonate + oral fluids

We repeated the same 3 threshold analyses in the emergency population to compare sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids. When the baseline risk of CI-AKI was varied between extreme values (1% and 50%; base case 19.6%) the INMB becomes negative at risk values of around 17% and below, indicating that the ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids exceeds £20,000 per QALY gained at these values (Figure 24). Notably, the 95% confidence interval for the baseline risk of CI-AKI in the emergency population is 10.6% to 33.4%, therefore the INMB becomes negative within the plausible boundaries for this parameter.

Figure 24. Threshold analysis – baseline risk of CI-AKI, emergency population, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids

When varying the mortality OR for CI-AKI versus no CI-AKI between extreme values (1 and 10; base case 5.59) to determine the effect on the ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids (Figure 25), the INMB becomes negative at OR values of around 3.25 or less, which is within the 95% confidence interval of the mortality OR (2.69 to 11.61).

Figure 25. Threshold analysis – mortality odds ratio for CI-AKI vs no CI-AKI, emergency population, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids

We varied the risk of ESRD following an episode of CI-AKI in the emergency population to determine whether it has the potential to meaningfully affect the ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids (Figure 26). We varied this parameter between extreme values (0% and 10%; base case 4.1%). Probabilities of approximately 3.25% and below result in a negative INMB.

Figure 26. Threshold analysis – probability of ESRD following CI-AKI, emergency population, sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids

Elective setting

We undertook a scenario analysis in which we assume that all patients are undergoing elective procedures. This involves using the synthesised baseline risk of CI-AKI from elective trials only (see Table 19), and the same assumptions surrounding excess bed day costs as the base-case analysis. This represents a population at comparatively low risk of CI-AKI. Base case cost–utility results are presented in Table 32. Compared with the base case ICER of £510,922 per QALY gained, the ICER increases to £655,323 per QALY gained in the elective population. All other interventions are dominated by oral fluids.

Table 32. Results – deterministic cost-effectiveness results, elective setting, without sodium bicarbonate + oral fluids

Figure 27 plots these results on the cost-effectiveness plane. As for the base case, the extremely steep gradient of the red line represents the very high ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids.

Figure 27. Cost-effectiveness plane, elective setting, without sodium bicarbonate + oral fluids

Figure 28 presents the CEAC. Results are similar to the base case in that oral fluids have the highest probability of cost effectiveness and have the highest expected net benefit across all QALY values.

Figure 28. Results – CEAC, elective setting, without sodium bicarbonate + oral fluids

The CEAC for the grouped regimens in the elective setting (Figure 29) is similar to the equivalent figure when elective and emergency are grouped (Figure 15), although the QALY value at which a sodium chloride 0.9% and/or sodium bicarbonate regimen overtakes oral fluids is higher in the elective setting (at approximately £27,000 per QALY).

Figure 29. Cost-effectiveness acceptability curve, without sodium bicarbonate + oral fluids, grouped regimens – elective

We also undertook an OSA for this scenario (Figure 30). Results are very similar to the analysis conducted for the base case without sodium bicarbonate (oral) + oral fluids removed (Figure 30), with only the relative treatment effects for oral fluids alone, or sodium chloride 0.9% (IV) + sodium bicarbonate (IV) having the potential to result in a positive incremental net monetary benefit.

Figure 30. Results – OSA, elective setting, without sodium bicarbonate + oral fluids

Discussion

Principal findings

The aim of the current analysis was to answer the research question ‘What is the comparative clinical and cost effectiveness of N-acetylcysteine (NAC) and/or fluids in preventing contrast induced acute kidney injury (CI-AKI) in at risk adults?’. To answer this question, we modelled a population with an average age of 70 years who have CKD Stages 3–5 (pre-RRT) and are at risk of CI-AKI from PCI procedures, using the results of the clinical NMA to inform the relative treatment effects for the different interventions.

In the base-case analysis, sodium bicarbonate (oral) plus oral fluids dominates all other interventions. Other than the effectiveness of the intervention (which is extremely uncertain), none of the parameters varied in sensitivity analyses change this result. Notably, the model results directly reflect results of the NMA, indicating that the probability of CI-AKI is the key parameter driving model results. When sodium bicarbonate (oral) plus oral fluids is removed from the decision space, oral fluids alone become the most cost-effective intervention. The only intervention not dominated by oral fluids is sodium chloride 0.9% (IV) + sodium bicarbonate (IV), with a base-case ICER of £510,922 per QALY gained. We conducted scenario analyses in emergency and elective populations with sodium bicarbonate (oral) plus oral fluids removed from the decision space. The ICER for sodium chloride 0.9% (IV) + sodium bicarbonate (IV) versus oral fluids increased further in the elective population to £655,323 per QALY gained; however, it dropped to £16,112 per QALY gained in the emergency population. All other interventions remain dominated in all scenarios.

We conducted an additional analysis separating results from the 17 strategies down into a simple 3-way split: (i) oral fluids alone, (ii) intravenous regimens with sodium chloride 0.9% and/or sodium bicarbonate (as currently recommended), (iii) other options (including oral NAC alone, no hydration regimen and IV sodium chloride 0.45%). Probabilistic results suggest that there is about a 60% chance that 1 or other of the simulated regimens containing intravenous sodium chloride 0.9% and/or intravenous sodium bicarbonate provides best value in the emergency setting (if a QALY is valued at £20,000). Oral fluids alone have the highest probability of being cost-effective at £20,000 per QALY in the grouped (emergency plus elective) and elective only subgroups.

Strengths of the analysis

Although this was an update of an existing model, we have made various changes and additions to use the most recent data available to us and refine the clinical pathway in line with the committee’s advice. As an example, the previous model did not incorporate the costs and QoL of kidney transplantation within the CKD stage 5 state, while we felt it was an important component and therefore included it within the updated model. We were also able to find multiple more recent or more appropriate sources of data for model parameterisation. For example, we obtained the risk of ESRD following CI-AKI from a study in people with CKD specifically (Chan et al., 2015), and we obtained the mortality relative risks for the RRT state from the UK Renal Registry (UK Renal Registry, 2018).

A key strength of the analysis is that it relies on an NMA for estimates of the relative treatment effects. To our knowledge, this is the most up-to-date estimate of the treatment effects for the included interventions. Furthermore, although the base-case results of the model directly reflect the results of the NMA, we synthesised a wealth of additional types of data from various sources to help the committee make informed recommendations that were not solely based on clinical effectiveness. As part of this, we presented long-term clinical outcomes over patients’ lifetimes to the committee, such as life-years and rates of ESRD.

The analyses presented here also benefit from extensive one-way and scenario analyses, as well as a PSA. All parameters and key scenarios were included in univariable analyses, and we explored key inputs in greater detail, including the use of alternative data sources. These were subject to different scenarios and threshold analyses. In particular, our modelling of emergency and elective subgroups showed the potential for important distinctions in cost–utility outcomes. In the emergency scenario, we were able to show how the balance of costs and benefits changes at different baseline risks of CI-AKI.

Lastly, the model was updated in close collaboration with the expert guideline committee. As part of this, the committee had several opportunities to review and discuss the model structure and inputs. This ensured the model had a high degree of external validity and was an appropriate representation of the true clinical pathway in CI-AKI.

Limitations of the analysis

Although we aimed to match the population from the clinical review as closely as possible, it was necessary to make some simplifying assumptions that meant there were some deviations from the original PICO in terms of clinical characteristics. For example, we only looked at people with stages 3–5 CKD (pre-dialysis) to represent those ‘at risk’ of CI-AKI, while the clinical review used a broader definition and therefore may have included people with, for example, myocardial infarction. Similarly, we used repeat PCI as a proxy for repeat scans, while the reasons for contrast administration within the clinical review varied.

There were some parameters for which suitable data could not be found. In such cases we asked the committee for their advice or used data sources that were not directly applicable to the population of interest. For example, we were unable to find an appropriate UK study that reports the utility value associated with CI-AKI in people with CKD; therefore, we used data in patients with AKI (from any cause) who were critically ill and only a fraction of whom had CKD (Nisula et al., 2013). An additional example is the number of temporary dialysis sessions required after CI-AKI; this did not appear to be reported in the literature, so we asked the committee for their advice on the number and type of dialysis sessions.

In some cases, we were able to find data, but it did not exactly match our needs or our population of interest. An example of this is mortality in CKD Stages 3–5: the mortality data for stage 3–5 CKD are taken from a cohort of people who have stage 3 CKD at baseline, and therefore may underestimate mortality. The mortality data for RRT, however, are taken from the UK renal registry, which is a good representation of our population of interest. In addition, there were some sources that did not report the data with the appropriate uncertainty estimates for the PSA. An example of this is the version of the NHS reference costs used within the model (2017–18) does not report the lower and upper quartiles for the cost estimates. We therefore assumed costs were fixed, which means uncertainty surrounding the reference costs is not accounted for within the model. Arguably, however, there is no parameter uncertainty attached to NHS reference costs, as they represent all NHS activity, and are, therefore, not subject to sampling error.

There are some attributes of the clinical and disease pathway that we decided not to model to avoid the model becoming overly complex. For example, people who have had an episode of CI-AKI and then recover are still likely to be at an increased risk of mortality and other long-term complications. This has not been incorporated within the model, and as such mortality rates in the Stage 3–5 group may be underestimated as the state will include some people who have had one or more cases of CI-AKI. Similarly, we assume that people who are within the Stage 3–5 state have the same probability of repeat scans, regardless of whether they have previously had an episode of CI-AKI. In clinical practice, it is likely that health professionals will be less likely to recommend scans in people who have had prior CI-AKI, particularly if alternatives are available.

Although a strength of the analysis was our scenario analysis in elective and emergency settings, there were additional subgroups specified in the review protocol that we did not have enough data to explore, for example people with diabetes, sepsis or hypovolaemia. This could be an area for future research.

The model results are based on results of the associated NMA, which has its own limitations that need to be considered when interpreting results of the model. When the model is analysed deterministically, only the point estimates from the NMA are used, which is why sodium bicarbonate (oral) plus oral fluids appears superior to all other options. However, data on the effectiveness of sodium bicarbonate (oral) plus oral fluids are limited and uncertain, and as such the point estimate has an extremely wide credible interval. The committee was aware of this and accounted for it during decision-making.

Comparison with other CUAs

No published cost–utility analyses were found to help answer the review question during the systematic literature search; therefore, there is a lack of clear reference point for this analysis. This was also the case during development of the previous version of the guideline.

Conclusions

In the base-case analysis, sodium bicarbonate (oral) plus oral fluids dominates all other interventions. However, the evidence surrounding the effectiveness of this intervention is extremely uncertain. Upon removal of this intervention from the decision space, oral fluids become the most cost effective; however, this is sensitive to assumptions surrounding the underlying risk of CI-AKI in the population. In patients who are undergoing elective procedures and are therefore at a lower baseline risk of CI-AKI, oral fluids remain the most cost effective. However, for patients undergoing emergency procedures and who are at a higher risk of CI-AKI, sodium chloride 0.9% plus sodium bicarbonate becomes cost-effective. When interventions are grouped, any regimen containing intravenous sodium chloride 0.9% and/or intravenous sodium bicarbonate provides best value in the emergency setting. These results indicate that an IV regimen including sodium chloride 0.9% and/or sodium bicarbonate is cost effective for people who are at a high risk of CI-AKI, while oral fluids may be sufficient for people at a lower risk.

Table of parameters

All parameters used in the model are summarised in Table 33, including details of the distributions and parameters used in probabilistic analysis. Dark grey shading indicates those parameters that were used in the previous version of the model and were replaced with updated parameters in the current base case.

Table 33. Full list of parameters used within the model

References

Brennan DC, Daller JA, Lake KD et al. (2006) Rabbit antithymocyte globulin versus basiliximab in renal transplantation. The New England journal of medicine 355(19): 1967–1977. [PubMed: 17093248]
Briguori C, Airoldi F, D’Andrea D et al. (2007) Renal Insufficiency Following Contrast Media Administration Trial (REMEDIAL): a randomized comparison of 3 preventive strategies. Circulation 115(10): 1211–1217. [PubMed: 17309916]
Briguori C, Manganelli F, Scarpato P et al. (2002) Acetylcysteine and contrast agent-associated nephrotoxicity. Journal of the American College of Cardiology 40(2): 298–303. [PubMed: 12106935]
Chan W, Ivanov J, Ko D et al. (2015) Clinical outcomes of treatment by percutaneous coronary intervention versus coronary artery bypass graft surgery in patients with chronic kidney disease undergoing index revascularization in Ontario. Circulation. Cardiovascular interventions 8(1). [PubMed: 25582144]
Cho R, Javed N, Traub D et al. (2010) Oral hydration and alkalinization is noninferior to intravenous therapy for prevention of contrast-induced nephropathy in patients with chronic kidney disease. Journal of interventional cardiology 23(5): 460–466. [PubMed: 20796166]
Commercial Medicines Unit. Drugs and pharmaceutical electronic market information tool (eMIT). Accessed at: https://www.gov.uk/government/publications/drugs-and-pharmaceutical-electronic-market-information-emit [Accessed May 2019].
Curtis L & Burns A. (2018). Unit cost of health and social care 2018. University of Kent, UK.
Dangas G, Iakovou I, Nikolsky E et al. (2005) Contrast-induced nephropathy after percutaneous coronary interventions in relation to chronic kidney disease and hemodynamic variables. The American journal of cardiology 95(1): 13–19. [PubMed: 15619387]
Department of Health and Social Care (2016). NHS reference costs 2015 to 2016. Accessed at: https://www.gov.uk/government/publications/nhs-reference-costs-2015-to-2016.
Eriksen BO and Ingebretsen OC (2006) The progression of chronic kidney disease: a 10-year population-based study of the effects of gender and age. Kidney international 69(2): 375–382. [PubMed: 16408129]
Hoste EA, Doom S, De Waele J et al. (2011) Epidemiology of contrast-associated acute kidney injury in ICU patients: a retrospective cohort analysis. Intensive care medicine 37(12): 1921–1931. [PubMed: 22048719]
Hussain JA; Mooney A; Russon L (2013) Comparison of survival analysis and palliative care involvement in patients aged over 70 years choosing conservative management or renal replacement therapy in advanced chronic kidney disease. Palliative medicine 27(9): 829–839. [PubMed: 23652841]
James MT, Ghali WA, Knudtson ML et al. (2011) Associations between acute kidney injury and cardiovascular and renal outcomes after coronary angiography. Circulation 123(4): 409–416. [PubMed: 21242477]
James MT, Hemmelgarn BR, Wiebe N et al. (2010) Glomerular filtration rate, proteinuria, and the incidence and consequences of acute kidney injury: a cohort study. Lancet (London, England) 376(9758): 2096–2103. [PubMed: 21094997]
Jesky MD, Dutton M, Dasgupta I et al. (2016) Health-Related Quality of Life Impacts Mortality but Not Progression to End-Stage Renal Disease in Pre-Dialysis Chronic Kidney Disease: A Prospective Observational Study. PloS one 11(11): e0165675. [PMC free article: PMC5104414] [PubMed: 27832126]
Joint Formulary Committee. British National Formulary (online) London: BMJ Group and Pharmaceutical Press <http://www.medicinescomplete.com= [Accessed June 2019].
Jones-Hughes T, Snowsill T, Haasova M et al. (2016) Immunosuppressive therapy for kidney transplantation in adults: a systematic review and economic model. Health technology assessment (Winchester, England) 20(62): 1–594. [PMC free article: PMC5018688] [PubMed: 27578428]
Kaltenthaler E, Tappenden P, Paisley S, Squires H. (2011). NICE DSU Technical Support Document 13: Identifying and reviewing evidence to inform the conceptualisation and population of cost-effectiveness models. Accessed at: http://www.nicedsu.org.uk. [PubMed: 28481494]
Kama A, Yılmaz S, Yaka E et al. (2014) Comparison of short-term infusion regimens of N-acetylcysteine plus intravenous fluids, sodium bicarbonate plus intravenous fluids, and intravenous fluids alone for prevention of contrast-induced nephropathy in the emergency department. Academic emergency medicine: official journal of the Society for Academic Emergency Medicine 21(6): 615–622. [PubMed: 25039544]
Kind P, Hardman G, Macran S. (1999) UK population norms for EQ-5D. CHE Discussion Paper 175, University of York, UK.
Liem YS; Bosch JL; Hunink MG (2008) Preference-based quality of life of patients on renal replacement therapy: a systematic review and meta-analysis. Value in health: the journal of the International Society for Pharmacoeconomics and Outcomes Research 11(4): 733–741. [PubMed: 18194399]
Maioli M, Toso A, Leoncini M et al. (2012) Persistent renal damage after contrast-induced acute kidney injury: incidence, evolution, risk factors, and prognosis. Circulation 125(25): 3099–3107. [PubMed: 22592896]
Maioli M, Toso A, Leoncini M et al. (2008) Sodium bicarbonate versus saline for the prevention of contrast-induced nephropathy in patients with renal dysfunction undergoing coronary angiography or intervention. Journal of the American College of Cardiology 52(8): 599–604. [PubMed: 18702961]
Marks A, Black C, Fluck N et al. (2012) Translating chronic kidney disease epidemiology into patient care--the individual/public health risk paradox. Nephrology, dialysis, transplantation: official publication of the European Dialysis and Transplant Association - European Renal Association: iii65. [PubMed: 22532617]
Martin-Moreno PL, Varo N, Martínez-Ansó E et al. (2015) Comparison of Intravenous and Oral Hydration in the Prevention of Contrast-Induced Acute Kidney Injury in Low-Risk Patients: A Randomized Trial. Nephron 131(1): 51–58. [PubMed: 26336919]
Mehran R, Aymong ED, Nikolsky E et al. (2004) A simple risk score for prediction of contrast-induced nephropathy after percutaneous coronary intervention: development and initial validation. Journal of the American College of Cardiology 44(7): 1393–1399. [PubMed: 15464318]
Mueller C, Buerkle G, Buettner HJ et al. (2002) Prevention of contrast media-associated nephropathy: randomized comparison of 2 hydration regimens in 1620 patients undergoing coronary angioplasty. Archives of internal medicine 162(3): 329–336. [PubMed: 11822926]
NHS Blood and Transplant (2018). Organ Donation and Transplantation Activity Report 2017/18. Accessed at: https://nhsbtdbe.blob.core.windows.net/umbracoassets/1848/transplant-activity-report-2017-2018.pdf.
NHS Business Services Authority (2019a). Drug Tariff, June 2019. Accessed at: https://www.nhsbsa.nhs.uk/pharmacies-gp-practices-and-appliance-contractors/drug-tariff.
NHS Business Services Authority (2019b). Prescription Cost Analysis (PCA) data, March 2019. https://www.nhsbsa.nhs.uk/prescription-data/dispensing-data/prescription-cost-analysis-pca-data.
NHS Improvement (2018). National schedule of reference costs 2017–18. Accessed at: https://improvement.nhs.uk/resources/reference-costs/#rc1718.
NICE (2018). Renal replacement therapy and conservative management. NICE guideline [NG107]. [PubMed: 32209597]
NICE (2015). Chronic kidney disease: managing anaemia. NICE guideline [NG8].
NICE (2013a). Acute kidney injury: prevention, detection and management. NICE clinical guideline [CG169].
NICE (2013b). Chronic kidney disease (stage 4 or 5): management of hyperphosphataemia. Clinical guideline [CG157].
Nisula S, Vaara ST, Kaukonen KM et al. (2013) Six-month survival and quality of life of intensive care patients with acute kidney injury. Critical care (London, England) 17(5): R250. [PMC free article: PMC4056803] [PubMed: 24148658]
Office for National Statistics (2018). National life tables, UK: 2015–2017. Accessed at: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/lifeexpectancies/datasets/nationallifetablesenglandreferencetables.
Rashid ST, Salman M, Myint F et al. (2004) Prevention of contrast-induced nephropathy in vascular patients undergoing angiography: a randomized controlled trial of intravenous N-acetylcysteine. Journal of vascular surgery 40(6): 1136–1141. [PubMed: 15622367]
Sadineni R, Karthik KR, Swarnalatha G et al. (2017) N-acetyl cysteine versus allopurinol in the prevention of contrast nephropathy in patients with chronic kidney disease: A randomized controlled trial. Indian journal of nephrology 27(2): 93–98. [PMC free article: PMC5358166] [PubMed: 28356658]
Serruys PW, Morice MC, Kappetein AP et al. (2009) Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease. The New England journal of medicine 360(10): 961–972. [PubMed: 19228612]
UK Renal Registry (2018) 20th Annual Report of the Renal Association. Nephron 139 (suppl1).
Villar E, Remontet L, Labeeuw M et al. (2007) Effect of age, gender, and diabetes on excess death in end-stage renal failure. Journal of the American Society of Nephrology: JASN 18(7): 2125–2134. [PubMed: 17582163]

Appendix M. Excluded studies

Clinical studies

Study	Reason
(2007) MEENA (A Randomized Controlled Trial for the Prevention of Contrast-induced Nephropathy with Sodium Bicarbonate in Persons Undergoing Coronary Angiography). Clinical Cardiology 30(8): 416–416	- Conference abstract
Agarwal, Shiv Kumar, Mohareb, Sameh, Patel, Achint et al. (2015) Systematic oral hydration with water is similar to parenteral hydration for prevention of contrast-induced nephropathy: an updated meta-analysis of randomised clinical data. Open heart 2(1): e000317 [PMC free article: PMC4600249] [PubMed: 26468404]	- More recent systematic review covers the same topic
Ahmed, K., McVeigh, T., Cerneviciute, R. et al. (2018) Effectiveness of contrast-associated acute kidney injury prevention methods; A systematic review and network meta-analysis. BMC Nephrology 19(1): 323 [PMC free article: PMC6234687] [PubMed: 30424723]	- Systematic review used as source of primary studies [Some of the interventions were not relevant for the update.]
Alessandri, N., Lanzi, L., Garante, C. M. et al. (2013) Prevention of acute renal failure post-contrast imaging in cardiology: a randomized study. European review for medical and pharmacological sciences 17suppl1: 13–21 [PubMed: 23436661]	- Not a relevant study design [Retrospective observational study]
Ali-Hasan-Al-Saegh, Sadeq, Mirhosseini, Seyed Jalil, Ghodratipour, Zahra et al. (2017) Strategies Preventing Contrast-Induced Nephropathy After Coronary Angiography: A Comprehensive Meta-Analysis and Systematic Review of 125 Randomized Controlled Trials. Angiology 68(5): 389–413 [PubMed: 27485363]	- More recent systematic review covers the same topic
Ali-Hassan-Sayegh, S., Mirhosseini, S. J., Rahimizadeh, E. et al. (2015) Current status of sodium bicarbonate in coronary angiography: An updated comprehensive meta-analysis and systematic review. Cardiology Research and Practice 2015: 690308 [PMC free article: PMC4417980] [PubMed: 25973282]	- More recent systematic review covers the same topic
Alonso, A., Lau, J., Jaber, B. L. et al. (2004) Prevention of Radiocontrast Nephropathy with N-Acetylcysteine in Patients with Chronic Kidney Disease: a Meta-Analysis of Randomized, Controlled Trials. American journal of kidney diseases 43(1): 1–9 [PubMed: 14712421]	- Does not contain a population of people at risk of CI-AKI
Alonso, Pau, Sanz, Jorge, Garcia-Orts, Ana et al. (2017) Usefulness of Sodium Bicarbonate for the Prevention of Contrast-Induced Nephropathy in Patients Undergoing Cardiac Resynchronization Therapy. The American journal of cardiology 120(9): 1584–1588 [PubMed: 28844518]	- Does not contain a population of people at risk of CI-AKI
Bagshaw, S. M. and Ghali, W. A. (2004) Acetylcysteine for prevention of contrast-induced nephropathy after intravascular angiography: a systematic review and meta-analysis. BMC medicine 2: 38 [PMC free article: PMC526263] [PubMed: 15500690]	- More recent systematic review covers the same topic
Bailey, Michael, McGuinness, Shay, Haase, Michael et al. (2015) Sodium bicarbonate and renal function after cardiac surgery: a prospectively planned individual patient meta-analysis. Anesthesiology 122(2): 294–306 [PubMed: 25501691]	- More recent systematic review covers the same topic
Balderramo D, Verdu M, Ramacciotti. C.F. et al. (2004) Renoprotective 30 effect of high periprocedural doses of oral N-acetylcysteine in patients scheduled to undergo a same-day angiography. Revista De La Facultad De Ciencias Medicas De Cordoba 61(2) [PubMed: 16211988]	- Does not contain a population of people at risk of CI-AKI
Berwanger, Otavio, Cavalcanti, Alexandre Biasi, Sousa, Amanda M. G. et al. (2013) Acetylcysteine for the prevention of renal outcomes in patients with diabetes mellitus undergoing coronary and peripheral vascular angiography: a substudy of the acetylcysteine for contrast-induced nephropathy trial. Circulation. Cardiovascular interventions 6(2): 139–45 [PubMed: 23572490]	- Secondary publication of an included study that does not provide any additional relevant information
Biernacka-Fialkowska, Barbara, Szuksztul, Marta, Suslik, Wojciech et al. (2018) Intravenous N-acetylcysteine for the PRevention Of Contrast-induced nephropathy - a prospective, single-center, randomized, placebo-controlled trial. The INPROC trial. Postepy w kardiologii interwencyjnej = Advances in interventional cardiology 14(1): 59–66 [PMC free article: PMC5939546] [PubMed: 29743905]	- Does not contain a population of people at risk of CI-AKI
Boccalandro, F., Amhad, M., Smalling, R. W. et al. (2003) Oral acetylcysteine does not protect renal function from moderate to high doses of intravenous radiographic contrast. Catheterization and cardiovascular interventions 58(3): 336–341 [PubMed: 12594698]	- Not a relevant study design
Brown, Jeremiah R., Pearlman, Daniel M., Marshall, Emily J. et al. (2016) Meta-Analysis of Individual Patient Data of Sodium Bicarbonate and Sodium Chloride for All-Cause Mortality After Coronary Angiography. The American journal of cardiology 118(10): 1473–1479 [PMC free article: PMC6579735] [PubMed: 27642111]	- More recent systematic review covers the same topic
Brueck, Martin, Cengiz, Huelya, Hoeltgen, Reinhard et al. (2013) Usefulness of N-acetylcysteine or ascorbic acid versus placebo to prevent contrast-induced acute kidney injury in patients undergoing elective cardiac catheterization: a single-center, prospective, randomized, double-blind, placebo-controlled trial. The Journal of invasive cardiology 25(6): 276–83 [PubMed: 23735352]	- Does not contain a population of people at risk of CI-AKI
Burns KE; Priestap F; Martin C (2010) N-acetylcysteine in critically ill patients undergoing contrast-enhanced computed tomography: a randomized trial.. Clinical nephrology 74(4): 323–326 [PubMed: 20875388]	- Letter to editor
Busch, Sarah Victoria Ekelof, Jensen, Svend Eggert, Rosenberg, Jacob et al. (2013) Prevention of contrast-induced nephropathy in STEMI patients undergoing primary percutaneous coronary intervention: a systematic review. Journal of interventional cardiology 26(1): 97–105 [PubMed: 23240788]	- More recent systematic review covers the same topic
Chen, H. W., Zhang, J. J., Xiong, D. et al. (2016) Prevention and Treatment of Shenkang Injection for Contrast-induced Nephropathy in Elder Patients with Chronic Kidney Disease. Zhongguo zhong xi yi jie he za zhi zhongguo zhongxiyi jiehe zazhi = chinese journal of integrated traditional and western medicine 36(7): 792–796 [PubMed: 30634203]	- Study not reported in English [Chinese]
Cheungpasitporn, Wisit, Thongprayoon, Charat, Brabec, Brady A. et al. (2014) Oral hydration for prevention of contrast-induced acute kidney injury in elective radiological procedures: a systematic review and meta-analysis of randomized controlled trials. North American journal of medical sciences 6(12): 618–24 [PMC free article: PMC4290050] [PubMed: 25599049]	- More recent systematic review covers the same topic
Dabare, Dilan, Banihani, Mohammed, Gibbs, Paul et al. (2013) Does bicarbonate prevent contrast-induced nephropathy in cardiovascular patients undergoing contrast imaging?. Interactive cardiovascular and thoracic surgery 17(6): 1028–35 [PMC free article: PMC3829505] [PubMed: 23996732]	- More recent systematic review covers the same topic
Ding, L.; Zhuang, G. H.; Ding, B. (2016) Clinical application of intravenous hydration or oral hydration in preventing contrast-induced nephropathy in patients with cardiac insufficiency. Journal of interventional radiology (china) 25(1): 15–18	- Study not reported in English [Chinese]
Dong, Yuhao, Zhang, Bin, Liang, Long et al. (2016) How Strong Is the Evidence for Sodium Bicarbonate to Prevent Contrast-Induced Acute Kidney Injury After Coronary Angiography and Percutaneous Coronary Intervention?. Medicine 95(7): e2715 [PMC free article: PMC4998610] [PubMed: 26886610]	- More recent systematic review covers the same topic
Eskandarian, R., Yarmohamadi, M., Zaker-Tavalae, M. et al. (2018) The standard dose versus double dose of n-acetylcysteine to prevent contrast-induced nephropathy; a randomized controlled clinical trial. Journal of Nephropathology 7(3): 145–150	- Data not reported in an extractable format
Giacoppo, Daniele, Gargiulo, Giuseppe, Buccheri, Sergio et al. (2017) Preventive Strategies for Contrast-Induced Acute Kidney Injury in Patients Undergoing Percutaneous Coronary Procedures: Evidence From a Hierarchical Bayesian Network Meta-Analysis of 124 Trials and 28 240 Patients. Circulation. Cardiovascular interventions 10(5) [PubMed: 28487354]	- Systematic review used as source of primary studies [Some of the interventions were not relevant for the update. The NMA was only for patients undergoing percutaneous coronary procedures.]
Glaza, M.; Rutkowski, B.; Szolkiewicz, M. (2018) Prevention of contrast-induced nephropathy in patients after percutaneous coronary intervention: A single-center prospective study. Clinical Nephrology 90(5): 370–372 [PubMed: 30049301]	- Letter to editor [About PRESERVE trial]
Guru, V. and Fremes, S. E. (2004) The role of N-acetylcysteine in preventing radiographic contrast-induced nephropathy. Clinical nephrology 62(2): 77–83 [PubMed: 15356963]	- More recent systematic review covers the same topic
Heguilen, Ricardo M., Liste, Amador A., Payaslian, Miguel et al. (2013) N-acethyl-cysteine reduces the occurrence of contrast-induced acute kidney injury in patients with renal dysfunction: a single-center randomized controlled trial. Clinical and experimental nephrology 17(3): 396–404 [PubMed: 23138396]	- Does not contain a population of people at risk of CI-AKI
Hiremath, Swapnil, Akbari, Ayub, Shabana, Wael et al. (2013) Prevention of contrast-induced acute kidney injury: is simple oral hydration similar to intravenous? A systematic review of the evidence. PloS one 8(3): e60009 [PMC free article: PMC3608617] [PubMed: 23555863]	- More recent systematic review covers the same topic
Inda-Filho, Antonio Jose, Caixeta, Adriano, Manggini, Marcia et al. (2014) Do intravenous N-acetylcysteine and sodium bicarbonate prevent high osmolal contrast-induced acute kidney injury? A randomized controlled trial. PloS one 9(9): e107602 [PMC free article: PMC4177831] [PubMed: 25254489]	- Does not contain a population of people at risk of CI-AKI
Izcovich, Ariel and Rada, Gabriel (2015) Should acetylcysteine be used to prevent contrast induced nephropathy?. Medwave 15(3): e6122 [PubMed: 25918998]	- More recent systematic review covers the same topic
Jang, Jae-Sik, Jin, Han-Young, Seo, Jeong-Sook et al. (2012) Sodium bicarbonate therapy for the prevention of contrast-induced acute kidney injury - a systematic review and meta-analysis. Circulation journal : official journal of the Japanese Circulation Society 76(9): 2255–65 [PubMed: 22975638]	- More recent systematic review covers the same topic
Jiang, Yufeng, Chen, Min, Zhang, Yiqing et al. (2017) Meta-analysis of prophylactic hydration versus no hydration on contrast-induced acute kidney injury. Coronary artery disease 28(8): 649–657 [PubMed: 28692484]	- More recent systematic review covers the same topic
Jurado-Roman, Alfonso, Hernandez-Hernandez, Felipe, Garcia-Tejada, Julio et al. (2015) Role of hydration in contrast-induced nephropathy in patients who underwent primary percutaneous coronary intervention. The American journal of cardiology 115(9): 1174–8 [PubMed: 25759106]	- Does not contain a population of people at risk of CI-AKI
Kanbay, M., Covic, A., Coca, S. G. et al. (2009) Sodium bicarbonate for the prevention of contrast-induced nephropathy: a meta-analysis of 17 randomized trials. International urology and nephrology 41(3): 617–627 [PubMed: 19396567]	- More recent systematic review covers the same topic
Kang, Xin, Hu, Da-Yong, Li, Chang-Bin et al. (2015) N-acetylcysteine for the prevention of contrast-induced nephropathy in patients with pre-existing renal insufficiency or diabetes: a systematic review and meta-analysis. Renal failure 37(10): 297–303 [PubMed: 26458505]	- More recent systematic review covers the same topic
Khaledifar, A., Momeni, A., Ebrahimi, A. et al. (2015) Comparison of N-acetylcysteine, ascorbic acid, and normal saline effect in prevention of contrast-induced nephropathy. ARYA Atherosclerosis 11(4) [PMC free article: PMC4593658] [PubMed: 26478730]	- Ocurrence of contrast induced AKI was not reported
Khan, Safi U., Khan, Muhammad U., Rahman, Hammad et al. (2019) A Bayesian network meta-analysis of preventive strategies for contrast-induced nephropathy after cardiac catheterization. Cardiovascular revascularization medicine : including molecular interventions 20(1): 29–37 [PubMed: 30757995]	- Systematic review used as source of primary studies [Some of the interventions were not relevant for the update.]
Kim, Byung Jin, Sung, Ki Chul, Kim, Bum Soo et al. (2010) Effect of N-Acetylcysteine on cystatin C-Based renaL function after Elective coronary angiography (ENABLE Study): A prospective, randomized trial. International Journal of Cardiology 138(3): 239–245 [PubMed: 18793808]	- Does not contain a population of people at risk of CI-AKI
Kimmel, Martin, Butscheid, Moritz, Brenner, Stefanie et al. (2008) Improved estimation of glomerular filtration rate by serum cystatin C in preventing contrast induced nephropathy by N-acetylcysteine or zinc?preliminary results. ndt 23(4): 1241–1245 [PubMed: 18174269]	- Study was halted before it finished recruiting
Koc, Fatih, Ozdemir, Kurtulus, Altunkas, Fatih et al. (2013) Sodium bicarbonate versus isotonic saline for the prevention of contrast-induced nephropathy in patients with diabetes mellitus undergoing coronary angiography and/or intervention: a multicenter prospective randomized study. Journal of investigative medicine : the official publication of the American Federation for Clinical Research 61(5): 872–7 [PubMed: 23552179]	- Does not contain a population of people at risk of CI-AKI
Kumar, A., Bhawani, G., Kumari, N. et al. (2014) Comparative study of renal protective effects of allopurinol and n-acetyl-cysteine on contrast induced nephropathy in patients undergoing cardiac catheterization. Journal of Clinical and Diagnostic Research 8(12): HC03–HC07 [PMC free article: PMC4316271] [PubMed: 25653965]	- Does not contain a population of people at risk of CI-AKI
Li, J., Jin, E., Yu, L. et al. (2017) Oral N-acetylcysteine for prophylaxis of contrast-induced nephropathy in patients following coronary angioplasty: A meta-analysis. Experimental and Therapeutic Medicine 14(2): 1568–1576 [PMC free article: PMC5525578] [PubMed: 28810622]	- More recent systematic review covers the same topic
Liu, R., Nair, D., Ix, J. et al. (2005) N-acetylcysteine for the prevention of contrast-induced nephropathy: a systematic review and meta-analysis. Journal of General Internal Medicine 20(2): 193–200 [PMC free article: PMC1490056] [PubMed: 15836554]	- More recent systematic review covers the same topic
Loomba, R. S., Shah, P. H., Aggarwal, S. et al. (2013) Role of N-acetylcysteine to prevent contrast-induced nephropathy: a meta-analysis. American Journal of Therapeutics: epub [PubMed: 23982694]	- More recent systematic review covers the same topic
Loomba, Rohit S., Shah, Parinda H., Aggarwal, Saurabh et al. (2016) Role of N-Acetylcysteine to Prevent Contrast-Induced Nephropathy: A Meta-analysis. American journal of therapeutics 23(1): e172–83 [PubMed: 23982694]	- More recent systematic review covers the same topic
Luo, Yu, Wang, Xiaodong, Ye, Zi et al. (2014) Remedial hydration reduces the incidence of contrast-induced nephropathy and short-term adverse events in patients with ST-segment elevation myocardial infarction: a single-center, randomized trial. Internal medicine (Tokyo, Japan) 53(20): 2265–72 [PubMed: 25318787]	- Does not contain a population of people at risk of CI-AKI
Ma, W. Q., Zhao, Y., Wang, Y. et al. (2018) Comparative efficacy of pharmacological interventions for contrast-induced nephropathy prevention after coronary angiography: a network meta-analysis from randomized trials. International Urology and Nephrology 50(6): 1085–1095 [PubMed: 29404930]	- Systematic review used as source of primary studies [Some of the interventions were not relevant for the update.]
Mahmoodi, Khalil, Sohrabi, Bahram, Ilkhchooyi, Farzad et al. (2014) The Efficacy of Hydration with Normal Saline Versus Hydration with Sodium Bicarbonate in the Prevention of Contrast-induced Nephropathy. Heart views : the official journal of the Gulf Heart Association 15(2): 33–6 [PMC free article: PMC4124663] [PubMed: 25104980]	- Does not contain a population of people at risk of CI-AKI
Manari, Antonio, Magnavacchi, Paolo, Puggioni, Enrico et al. (2014) Acute kidney injury after primary angioplasty: effect of different hydration treatments. Journal of cardiovascular medicine (Hagerstown, Md.) 15(1): 60–7 [PubMed: 24500238]	- Does not contain a population of people at risk of CI-AKI
Navarese, Eliano P., Gurbel, Paul A., Andreotti, Felicita et al. (2017) Prevention of contrast-induced acute kidney injury in patients undergoing cardiovascular procedures-a systematic review and network meta-analysis. PloS one 12(2): e0168726 [PMC free article: PMC5289438] [PubMed: 28151965]	- Systematic review used as source of primary studies [Some interventions were not relevant for this update. The NMA was only for patients undergoing cardiovascular procedures.]
O’Sullivan, S., Healy, D. A., Moloney, Mary Clarke et al. (2013) The role of N-acetylcysteine in the prevention of contrast-induced nephropathy in patients undergoing peripheral angiography: a structured review and meta-analysis. Angiology 64(8): 576–82 [PubMed: 23188834]	- More recent systematic review covers the same topic
Pakfetrat M, Nikoo MH, Malekmakan L et al. (2009) A comparison of sodium bicarbonate infusion versus normal saline infusion and its combination with oral acetazolamide for prevention of contrast-induced nephropathy: a randomized, double-blind trial.. International urology and nephrology 41(3): 629–634 [PubMed: 19137409]	- Does not contain a population of people at risk of CI-AKI
Pakfetrat, Maryam, Malekmakan, Leila, Salmanpour, Zahra et al. (2019) Comparison of Normal Saline, Ringer’s Lactate, and Sodium Bicarbonate for Prevention of Contrast-induced Nephropathy in Patients with Coronary Angiography: A Randomized Double-blind Clinical Trial. Indian journal of nephrology 29(1): 22–27 [PMC free article: PMC6375023] [PubMed: 30814789]	- Does not contain a population of people at risk of CI-AKI
Pandya, B., Chaloub, J., Parikh, V. et al. (2017) Contrast media use in patients with chronic kidney disease undergoing coronary angiography: A systematic review and meta-analysis of randomized trials. International Journal of Cardiology 228: 137–144 [PubMed: 27863354]	- More recent systematic review covers the same topic
Pezeshgi, Aiyoub, Parsamanesh, Negin, Farhood, Goodarz et al. (2015) Evaluation of the protective effect of N-acetylcysteine on contrast media nephropathy. Journal of renal injury prevention 4(4): 109–12 [PMC free article: PMC4685979] [PubMed: 26693496]	- Does not contain a population of people at risk of CI-AKI
Poletti, Pierre-Alexandre, Platon, Alexandra, De Seigneux, Sophie et al. (2013) N-acetylcysteine does not prevent contrast nephropathy in patients with renal impairment undergoing emergency CT: a randomized study. BMC nephrology 14: 119 [PMC free article: PMC3682900] [PubMed: 23731573]	- Data not reported in an extractable format [Ocurrence of CI-AKI <5 days was not reported]
Ratcliffe JA, Thiagarajah P, Chen J et al. (2009) Prevention of contrast-induced nephropathy: A randomized controlled trial of sodium bicarbonate and N-acetylcysteine.. The International journal of angiology : official publication of the International College of Angiology, Inc 18(4): 193–197 [PMC free article: PMC2903033] [PubMed: 22477552]	- Does not contain a population of people at risk of CI-AKI
Sadat U, Walsh SR, Norden AG et al. (2011) Does oral N-acetylcysteine reduce contrast-induced renal injury in patients with peripheral arterial disease undergoing peripheral angiography? A randomized-controlled study.. Angiology 62(3): 225–230 [PubMed: 20682612]	- Does not contain a population of people at risk of CI-AKI
Sar F, Saler T, Ecebay A et al. (2010) The efficacy of n-acetylcysteine in preventing contrast-induced nephropathy in type 2 diabetic patients without nephropathy.. Journal of nephrology 23(4): 478–482 [PubMed: 20383874]	- Does not contain a population of people at risk of CI-AKI
Sharp, Alexander J., Patel, Nishith, Reeves, Barney C. et al. (2019) Pharmacological interventions for the prevention of contrast-induced acute kidney injury in high-risk adult patients undergoing coronary angiography: a systematic review and meta-analysis of randomised controlled trials. Open heart 6(1): e000864 [PMC free article: PMC6350720] [PubMed: 30774964]	- More recent systematic review covers the same topic
Silva RG, Silva NG, Lucchesi F et al. (2010) Prevention of contrast-induced nephropathy by use of bicarbonate solution: preliminary results and literature review.. Jornal brasileiro de nefrologia : ‘orgao oficial de Sociedades Brasileira e Latino-Americana de Nefrologia 32(3): 292–302 [PubMed: 21103694]	- Data not reported in an extractable format [number of pts per group not reported (results [preliminary] reported on pages 293–294)]
Sinert, R. and Doty, C. I. (2009) Update: prevention of contrast-induced nephropathy in the emergency department. Annals of Emergency Medicine 54(1): e1–e5 [PubMed: 18926598]	- More recent systematic review covers the same topic
Su, Xiaole, Xie, Xinfang, Liu, Lijun et al. (2017) Comparative Effectiveness of 12 Treatment Strategies for Preventing Contrast-Induced Acute Kidney Injury: A Systematic Review and Bayesian Network Meta-analysis. American journal of kidney diseases : the official journal of the National Kidney Foundation 69(1): 69–77 [PubMed: 27707552]	- Systematic review used as source of primary studies [Some of the interventions are not relevant for this update.]
Subramaniam, Rathan M., Suarez-Cuervo, Catalina, Wilson, Renee F. et al. (2016) Effectiveness of Prevention Strategies for Contrast-Induced Nephropathy: A Systematic Review and Meta-analysis. Annals of internal medicine 164(6): 406–16 [PubMed: 26830221]	- More recent systematic review covers the same topic
Subramaniam, Rathan M., Wilson, Renee F., Turban, Sharon et al. (2016) Contrast-Induced Nephropathy: Comparative Effectiveness of Preventive Measures. Agency for Healthcare Research and Quality (US). AHRQ Comparative Effectiveness Reviews, Report No.: 15(16)-EHC023-EF [PubMed: 26866209]	- More recent systematic review covers the same topic
Sun, Zikai, Fu, Qiang, Cao, Longxing et al. (2013) Intravenous N-acetylcysteine for prevention of contrast-induced nephropathy: a meta-analysis of randomized, controlled trials. PloS one 8(1): e55124 [PMC free article: PMC3559541] [PubMed: 23383076]	- More recent systematic review covers the same topic
Tanaka A, Suzuki Y, Suzuki N et al. (2011) Does N-acetylcysteine reduce the incidence of contrast-induced nephropathy and clinical events in patients undergoing primary angioplasty for acute myocardial infarction?. Internal medicine (Tokyo, Japan) 50(7): 673–677 [PubMed: 21467697]	- Does not contain a population of people at risk of CI-AKI
Thayssen, Per, Lassen, Jens Flensted, Jensen, Svend Eggert et al. (2014) Prevention of contrast-induced nephropathy with N-acetylcysteine or sodium bicarbonate in patients with ST-segment-myocardial infarction: a prospective, randomized, open-labeled trial. Circulation. Cardiovascular interventions 7(2): 216–24 [PubMed: 24714489]	- Does not contain a population of people at risk of CI-AKI
Trivedi HS, Moore H, Nasr S et al. (2003) A randomized prospective trial to assess the role of saline hydration on the development of contrast nephrotoxicity.. Nephron. Clinical practice 93(1): c29 [PubMed: 12411756]	- Does not contain a population of people at risk of CI-AKI
Valette, Xavier, Desmeulles, Isabelle, Savary, Benoit et al. (2017) Sodium Bicarbonate Versus Sodium Chloride for Preventing Contrast-Associated Acute Kidney Injury in Critically Ill Patients: A Randomized Controlled Trial. Critical care medicine 45(4): 637–644 [PubMed: 28181941]	- Does not contain a population of people at risk of CI-AKI
Wang, Nelson, Qian, Pierre, Kumar, Shejil et al. (2016) The effect of N-acetylcysteine on the incidence of contrast-induced kidney injury: A systematic review and trial sequential analysis. International journal of cardiology 209: 319–27 [PubMed: 26922293]	- More recent systematic review covers the same topic
Weisbord, Steven D., Gallagher, Martin, Kaufman, James et al. (2013) Prevention of contrast-induced AKI: a review of published trials and the design of the prevention of serious adverse events following angiography (PRESERVE) trial. Clinical journal of the American Society of Nephrology : CJASN 8(9): 1618–31 [PMC free article: PMC3805082] [PubMed: 23660180]	- More recent systematic review covers the same topic
Wu, Mei-Yi, Hsiang, Hui-Fen, Wong, Chung-Shun et al. (2013) The effectiveness of N-Acetylcysteine in preventing contrast-induced nephropathy in patients undergoing contrast-enhanced computed tomography: a meta-analysis of randomized controlled trials. International urology and nephrology 45(5): 1309–18 [PubMed: 23283594]	- More recent systematic review covers the same topic
Xu, Renfan, Tao, Anyu, Bai, Yang et al. (2016) Effectiveness of N-Acetylcysteine for the Prevention of Contrast-Induced Nephropathy: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Journal of the American Heart Association 5(9) [PMC free article: PMC5079043] [PubMed: 27663415]	- More recent systematic review covers the same topic
Yang, Kun, Liu, Wenxian, Ren, Wei et al. (2014) Different interventions in preventing contrast-induced nephropathy after percutaneous coronary intervention. International urology and nephrology 46(9): 1801–7 [PubMed: 24966097]	- Does not contain a population of people at risk of CI-AKI
Yeganehkhah, Mohammad Reza, Iranirad, Leili, Dorri, Farshad et al. (2014) Comparison between three supportive treatments for prevention of contrast-induced nephropathy in high-risk patients undergoing coronary angiography. Saudi journal of kidney diseases and transplantation : an official publication of the Saudi Center for Organ Transplantation, Saudi Arabia 25(6): 1217–23 [PubMed: 25394438]	- Data not reported in an extractable format [Figure 1 and data reported in text (page 1219) don’t match]
Zagler, Axel, Azadpour, Maziar, Mercado, Carlos et al. (2006) N-acetylcysteine and contrast-induced nephropathy: a meta-analysis of 13 randomized trials. American heart journal 151(1): 140–5 [PubMed: 16368307]	- More recent systematic review covers the same topic
Zapata-Chica, Carlos Andres, Bello Marquez, Diana, Serna-Higuita, Lina Maria et al. (2015) Sodium bicarbonate versus isotonic saline solution to prevent contrast-induced nephropathy : a systematic review and meta-analysis. Colombia medica (Cali, Colombia) 46(3): 90–103 [PMC free article: PMC4640430] [PubMed: 26600623]	- More recent systematic review covers the same topic
Zhang, Bin, Liang, Long, Chen, Wenbo et al. (2015) The efficacy of sodium bicarbonate in preventing contrast-induced nephropathy in patients with pre-existing renal insufficiency: a meta-analysis. BMJ open 5(3): e006989 [PMC free article: PMC4368906] [PubMed: 25783425]	- More recent systematic review covers the same topic
Zhao, Shi-Jie, Zhong, Zhao-Shuang, Qi, Guo-Xian et al. (2016) The efficacy of N-acetylcysteine plus sodium bicarbonate in the prevention of contrast-induced nephropathy after cardiac catheterization and percutaneous coronary intervention: A meta-analysis of randomized controlled trials. International journal of cardiology 221: 251–9 [PubMed: 27404685]	- More recent systematic review covers the same topic
Zoungas, S., Ninomiya, T., Huxley, R. et al. (2009) Systematic review: sodium bicarbonate treatment regimens for the prevention of contrast-induced nephropathy. Annals of internal medicine 151(9): 631–638 [PubMed: 19884624]	- More recent systematic review covers the same topic

Economic studies

Study	Reason
Kooiman, de Vries, Van der Heyden, Sijpkens, van Dijkman, Wever et al. (2018) Randomized trial of one-hour sodium bicarbonate vs standard periprocedural saline hydration in chronic kidney disease patients undergoing cardiovascular contrast procedures. PloS one 13(2): e0189372. [PMC free article: PMC5805164] [PubMed: 29420536]	Does not include quality of life data.
Kooiman, Sijpkens, de Vries, Brulez, Hamming, van der Molen et al. (2014) A randomized comparison of 1-h sodium bicarbonate hydration versus standard peri-procedural saline hydration in patients with chronic kidney disease undergoing intravenous contrast-enhanced computerized tomography. European Renal Association 29(5); 1029–36. [PubMed: 24578471]	Does not include quality of life data.
Kotlyar, Keogh, Thavapalachandran, Allada, Sharp, Dias et al. (2005) Prehydration alone is sufficient to prevent contrast-induced nephropathy after day-only angiography procedures-a randomised controlled trial. Heart, lung & circulation 14(4); 245–51. [PubMed: 16360994]	Not an economic evaluation.
Nijssen, Rennenberg, Nelemans, Essers, Janssen, Vermeeren et al. (2017) Prophylactic hydration to protect renal function from intravascular iodinated contrast material in patients at high risk of contrast-induced nephropathy (AMACING): a prospective, randomised, phase 3, controlled, open-label, non-inferiority trial. Lancet 389: 1312–22 [PubMed: 28233565]	Does not include quality of life data.

Appendix N. Research recommendations

Research recommendation 1

Potential criterion	Explanation
Importance to patients, service users or the population	An eGFR <40 ml/min/1.73 m² is associated with an increased risk of CI-AKI but the committee discussed that the risk might be different at other eGFR thresholds. The included RCTs in this update did not report any data on different eGFR thresholds. Therefore, the committee recommended that further research is needed to find out what eGFR thresholds are related to the risk of contrast-induced acute kidney injury.
Relevance to NICE guidance	Low-priority: it was possible to make recommendations based on the available evidence, but new evidence in this area has the potential to alter the recommendations substantially.
Current evidence base	The included RCTs in this update did not report any data on different eGFR thresholds.
Equality	No specific equality concerns are relevant to this research recommendation.
Feasibility	The committee noted that it might not be possible to do this research on people with very low eGFRs because they may be too high risk to be included in research studies. However, it agreed that better evidence of risk stratification for CI-AKI in people with higher eGFRs would still improve clinical practice and patient safety

Question	Can risk of contrast-induced acute kidney injury (CI-AKI) be stratified by eGFR thresholds?
Population	Adults (18 and older) who are at risk (as defined by the study author) of contrast induced AKI.
Prognostic factor	eGFR thresholds
Outcomes	Occurrence of contrast-induced acute kidney injury
Study design	Prospective cohort studies.

Research recommendation 2

Question	What is the relative effectiveness and cost effectiveness of different oral fluids and different oral fluid regimes, both with and without oral NAC, at preventing CI-AKI?
Population	Adults (18 and older) who are at risk (as defined by the study author) of contrast induced AKI who are having oral rather than IV hydration to prevent CI-AKI.
Intervention	Different oral hydration regimes and agents, with or without NAC, including

Potential criterion	Explanation
Importance to patients, service users or the population	The committee agreed that oral hydration regimes were non-inferior to IV hydration regimes at preventing CI-AKI, however there was not enough comparative data to enable them to be clear about which oral fluid (if any) was most effective. They noted that it might not be possible to do this research on people with very low eGFRs because they may be too high risk to be included in research studies.
Relevance to NICE guidance	Low-priority: it was possible to make recommendations based on the available evidence, but new evidence in this area has the potential to alter the recommendations substantially.
Current evidence base	Only one study was identified that partly addressed this research question comparting oral fluids +bicarbonate to oral fluids alone (Cho, 2010).
Equality	No specific equality concerns are relevant to this research recommendation.
Feasibility	No feasibility concerns were identified.
Study design	Randomised controlled trials

Question	What is the relative effectiveness and cost effectiveness of different oral fluids and different oral fluid regimes, both with and without oral NAC, at preventing CI-AKI?
	Water Sodium bicarbonate Sodium citrate
Comparator	Each other or no oral or hydration
Outcomes	Occurrence of contrast-induced acute kidney injury (within 72 hours of contrast administration)

Appendix O. References

Clinical studies

Included studies

Adolph, Esther, Holdt-Lehmann, Birgit, Chatterjee, Tushar et al. (2008) Renal Insufficiency Following Radiocontrast Exposure Trial (REINFORCE): a randomized comparison of sodium bicarbonate versus sodium chloride hydration for the prevention of contrast-induced nephropathy. Coronary artery disease 19(6): 413–9 [PubMed: 18955835]
Agrawal, M., Wodlinger, A.M., Huggins, C.E. et al. (2004) Effect of N-Acetylcysteine on Serum Creatinine Concentration in Patients with Chronic Renal Insufficiency who Are Undergoing Coronary Angiography. Heart Drug 4(2): 87–91
Akyuz, Sukru, Karaca, Mehmet, Kemaloglu Oz, Tugba et al. (2014) Efficacy of oral hydration in the prevention of contrast-induced acute kidney injury in patients undergoing coronary angiography or intervention. Nephron. Clinical practice 128(12): 95–100 [PubMed: 25378376]
Albabtain, Monirah A., Almasood, Ali, Alshurafah, Hytham et al. (2013) Efficacy of ascorbic acid, N-acetylcysteine, or combination of both on top of saline hydration versus saline hydration alone on prevention of contrast-Induced nephropathy: a prospective randomized study. Journal of interventional cardiology 26(1): 90–6 [PubMed: 22994682]
Allaqaband S, Tumuluri R, Malik AM et al. (2002) Prospective randomized study of N-acetylcysteine, fenoldopam, and saline for prevention of radiocontrast-induced nephropathy.. Catheterization and cardiovascular interventions : official journal of the Society for Cardiac Angiography & Interventions 57(3): 279–283 [PubMed: 12410497]
Aslanger, E., Uslu, B., Akdeniz, C. et al. (2012) Intrarenal application of N-acetylcysteine for the prevention of contrast medium-induced nephropathy in primary angioplasty. Coronary artery disease 23(4): 265–270 [PubMed: 22343798]
Baskurt, M., Okcun, B., Abaci, O. et al. (2009) N-acetylcysteine versus N-acetylcysteine + theophylline for the prevention of contrast nephropathy. European Journal of Clinical Investigation 39(9): 793–799 [PubMed: 19500141]
Berwanger, O. (2011) Acetylcysteine for prevention of renal outcomes in patients undergoing coronary and peripheral vascular angiography: main results from the randomized acetylcysteine for contrast-induced nephropathy trial (ACT). Circulation 124(11): 1250–1259 [PubMed: 21859972]
Boucek, Petr, Havrdova, Terezia, Oliyarnyk, Olena et al. (2013) Prevention of contrast-induced nephropathy in diabetic patients with impaired renal function: a randomized, double blind trial of sodium bicarbonate versus sodium chloride-based hydration. Diabetes research and clinical practice 101(3): 303–8 [PubMed: 23835495]
Brar SS, Shen AY, Jorgensen MB et al. (2008) Sodium bicarbonate vs sodium chloride for the prevention of contrast medium-induced nephropathy in patients undergoing coronary angiography: a randomized trial.. JAMA 300(9): 1038–1046 [PubMed: 18768415]
Briguori C, Airoldi F, D’Andrea D et al. (2007) Renal Insufficiency Following Contrast Media Administration Trial (REMEDIAL): a randomized comparison of 3 preventive strategies.. Circulation 115(10): 1211–1217 [PubMed: 17309916]
Briguori C, Manganelli F, Scarpato P et al. (2002) Acetylcysteine and contrast agent-associated nephrotoxicity.. Journal of the American College of Cardiology 40(2): 298–303 [PubMed: 12106935]
Caglar, I. M., Caglar, F. N. T., Conkbayir, C. et al. (2014) Contrast study: Comparision of nephroprotective three protocols: Acetylcysteine-sodium bicarbonate-theophylline, to prevent contrast-induced nephropathy. Russian Journal of Cardiology 105(1): 27–31
Carbonell N, Sanjuán R, Blasco M et al. (2010) N-acetylcysteine: short-term clinical benefits after coronary angiography in high-risk renal patients.. Revista espanola de cardiologia 63(1): 12–19 [PubMed: 20089221]
Carbonell, Nieves, Blasco, Marisa, Sanjuán, Rafael et al. (2007) Intravenous N-acetylcysteine for preventing contrast-induced nephropathy: A randomised trial. International Journal of Cardiology 115(1): 57–62 [PubMed: 16814414]
Castini, Diego, Lucreziotti, Stefano, Bosotti, Laura et al. (2010) Prevention of Contrast-induced Nephropathy: A Single Center Randomized Study. Clinical Cardiology 33(3): e63–e68 [PMC free article: PMC6653091] [PubMed: 20127900]
Chen, Shao Liang, Zhang, Junjie, Yei, Fei et al. (2008) Clinical outcomes of contrast-induced nephropathy in patients undergoing percutaneous coronary intervention: a prospective, multicenter, randomized study to analyze the effect of hydration and acetylcysteine. International journal of cardiology 126(3): 407–13 [PubMed: 17651830]
Cho R, Javed N, Traub D et al. (2010) Oral hydration and alkalinization is noninferior to intravenous therapy for prevention of contrast-induced nephropathy in patients with chronic kidney disease.. Journal of interventional cardiology 23(5): 460–466 [PubMed: 20796166]
Chong, E., Poh, K. K., Lu, Q. et al. (2015) Comparison of combination therapy of high-dose oral N-acetylcysteine and intravenous sodium bicarbonate hydration with individual therapies in the reduction of Contrast-induced Nephropathy during Cardiac Catheterisation and Percutaneous Coronary Intervention (CONTRAST): A multi-centre, randomised, controlled trial. International Journal of Cardiology 201: 237–242 [PubMed: 26301645]
Durham JD, Caputo C, Dokko J et al. (2002) A randomized controlled trial of N-acetylcysteine to prevent contrast nephropathy in cardiac angiography.. Kidney international 62(6): 2202–2207 [PubMed: 12427146]
Erturk, Mehmet, Uslu, Nevzat, Gorgulu, Sevket et al. (2014) Does intravenous or oral high-dose N-acetylcysteine in addition to saline prevent contrast-induced nephropathy assessed by cystatin C?. Coronary artery disease 25(2): 111–7 [PubMed: 24365793]
Ferrario, Francesca, Barone, Maria Teresa, Landoni, Giovanni et al. (2009) Acetylcysteine and non-ionic isosmolar contrast-induced nephropathy?a randomized controlled study. ndt 24(10): 3103–3107 [PubMed: 19549691]
Fung JW, Szeto CC, Chan WW et al. (2004) Effect of N-acetylcysteine for prevention of contrast nephropathy in patients with moderate to severe renal insufficiency: a randomized trial.. American journal of kidney diseases : the official journal of the National Kidney Foundation 43(5): 801–808 [PubMed: 15112170]
Goldenberg I, Shechter M, Matetzky S et al. (2004) Oral acetylcysteine as an adjunct to saline hydration for the prevention of contrast-induced nephropathy following coronary angiography. A randomized controlled trial and review of the current literature.. European heart journal 25(3): 212–218 [PubMed: 14972421]
Gomes, V O, Poli de Figueredo, C E, Caramori, P et al. (2005) <em>N</em>-acetylcysteine does not prevent contrast induced nephropathy after cardiac catheterisation with an ionic low osmolality contrast medium: a multicentre clinical trial. Heart 91(6): 774 [PMC free article: PMC1768952] [PubMed: 15894775]
Habib, Mohammed; Hillis, Alaa; Hammad, Amen (2016) N-acetylcysteine and/or ascorbic acid versus placebo to prevent contrast-induced nephropathy in patients undergoing elective cardiac catheterization: The NAPCIN trial; A single-center, prospective, randomized trial. Saudi journal of kidney diseases and transplantation : an official publication of the Saudi Center for Organ Transplantation, Saudi Arabia 27(1): 55–61 [PubMed: 26787567]
Hafiz, Abdul Moiz, Jan, M. Fuad, Mori, Naoyo et al. (2012) Prevention of contrast-induced acute kidney injury in patients with stable chronic renal disease undergoing elective percutaneous coronary and peripheral interventions: randomized comparison of two preventive strategies. Catheterization and cardiovascular interventions : official journal of the Society for Cardiac Angiography & Interventions 79(6): 929–37 [PubMed: 21542114]
Heng AE, Cellarier E, Aublet-Cuvelier B et al. (2008) Is treatment with N-acetylcysteine to prevent contrast-induced nephropathy when using bicarbonate hydration out of date?. Clinical nephrology 70(6): 475–484 [PubMed: 19049703]
Hsu C, Lee J, Lo P et al. (2007) Prevention of radiocontrast-induced nephropathy with N-acetylcysteine after cardiac angiography in diabetic patients with renal dysfunction. Mid-Taiwan Journal of Medicine 12(4)
Izani Wan Mohamed W and Darus, Z: Yusof Z (2008) Oral N-acetylcysteine in prevention of contrast induced nephropathy following coronary angiogram. International Medical Journal 15(5): 353–361
Jaffery, Z., Verma, A., White, C. J. et al. (2012) A randomized trial of intravenous n-acetylcysteine to prevent contrast induced nephropathy in acute coronary syndromes. Catheterization and cardiovascular interventions 79(6): 921–926 [PubMed: 21542122]
Kama, Ahmet, Yilmaz, Serkan, Yaka, Elif et al. (2014) Comparison of short-term infusion regimens of N-acetylcysteine plus intravenous fluids, sodium bicarbonate plus intravenous fluids, and intravenous fluids alone for prevention of contrast-induced nephropathy in the emergency department. Academic emergency medicine : official journal of the Society for Academic Emergency Medicine 21(6): 615–22 [PubMed: 25039544]
Kay J, Chow WH, Chan TM et al. (2003) Acetylcysteine for prevention of acute deterioration of renal function following elective coronary angiography and intervention: a randomized controlled trial.. JAMA 289(5): 553–558 [PubMed: 12578487]
Khalili H, Dashti-Khavidaki S, Tabifar H et al. (2006) N-acetylcysteine in the prevention of contrast agent-induced nephrotoxicity in patients undergoing computed tomography studies. Therapy 3(6)
Kitzler TM, Jaberi A, Sendlhofer G et al. (2012) Efficacy of vitamin E and N-acetylcysteine in the prevention of contrast induced kidney injury in patients with chronic kidney disease: a double blind, randomized controlled trial.. Wiener klinische Wochenschrift 124(910): 312–319 [PubMed: 22527829]
Koc, Fatih, Ozdemir, Kurtulus, Kaya, Mehmet Gungor et al. (2012) Intravenous N-acetylcysteine plus high-dose hydration versus high-dose hydration and standard hydration for the prevention of contrast-induced nephropathy: CASIS--a multicenter prospective controlled trial. International journal of cardiology 155(3): 418–23 [PubMed: 21106264]
Kooiman, J., Sijpkens, Y. W. J., van Buren, M. et al. (2014) Randomised trial of no hydration vs. sodium bicarbonate hydration in patients with chronic kidney disease undergoing acute computed tomography-pulmonary angiography. Journal of thrombosis and haemostasis : JTH 12(10): 1658–66 [PubMed: 25142085]
Kooiman, Judith, de Vries, Jean-Paul P. M., Van der Heyden, Jan et al. (2018) Randomized trial of one-hour sodium bicarbonate vs standard periprocedural saline hydration in chronic kidney disease patients undergoing cardiovascular contrast procedures. PloS one 13(2): e0189372 [PMC free article: PMC5805164] [PubMed: 29420536]
Kooiman, Judith, Sijpkens, Yvo W. J., de Vries, Jean-Paul P. M. et al. (2014) A randomized comparison of 1-h sodium bicarbonate hydration versus standard peri-procedural saline hydration in patients with chronic kidney disease undergoing intravenous contrast-enhanced computerized tomography. Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association 29(5): 1029–36 [PubMed: 24578471]
Kotlyar E, Keogh AM, Thavapalachandran S et al. (2005) Prehydration alone is sufficient to prevent contrast-induced nephropathy after day-only angiography procedures--a randomised controlled trial.. Heart, lung & circulation 14(4): 245–251 [PubMed: 16360994]
Lee, Seung-Whan, Kim, Won-Jang, Kim, Young-Hak et al. (2011) Preventive Strategies of Renal Insufficiency in Patients with Diabetes Undergoing Intervention or Arteriography (the PREVENT Trial). The American Journal of Cardiology 107(10): 1447–1452 [PubMed: 21420063]
MacNeill, Briain D., Harding, Scott A., Bazari, Hasan et al. (2003) Prophylaxis of contrast-induced nephropathy in patients undergoing coronary angiography. Catheterization and Cardiovascular Interventions 60(4): 458–461 [PubMed: 14624421]
Maioli, Mauro, Toso, Anna, Leoncini, Mario et al. (2008) Sodium Bicarbonate Versus Saline for the Prevention of Contrast-Induced Nephropathy in Patients with Renal Dysfunction Undergoing Coronary Angiography or Intervention. Journal of the American College of Cardiology 52(8): 599 [PubMed: 18702961]
Maioli, Mauro, Toso, Anna, Leoncini, Mario et al. (2011) Effects of hydration in contrast-induced acute kidney injury after primary angioplasty: a randomized, controlled trial. Circulation. Cardiovascular interventions 4(5): 456–62 [PubMed: 21972403]
Marenzi G, Assanelli E, Marana I et al. (2006) N-acetylcysteine and contrast-induced nephropathy in primary angioplasty.. The New England journal of medicine 354(26): 2773–2782 [PubMed: 16807414]
Martin-Moreno, Paloma L., Varo, Nerea, Martinez-Anso, Eduardo et al. (2015) Comparison of Intravenous and Oral Hydration in the Prevention of Contrast-Induced Acute Kidney Injury in Low-Risk Patients: A Randomized Trial. Nephron 131(1): 51–8 [PubMed: 26336919]
Masuda M, Yamada T, Mine T et al. (2007) Comparison of usefulness of sodium bicarbonate versus sodium chloride to prevent contrast-induced nephropathy in patients undergoing an emergent coronary procedure.. The American journal of cardiology 100(5): 781–786 [PubMed: 17719320]
Masuda M, Yamada T, Okuyama Y et al. (2008) Sodium bicarbonate improves long-term clinical outcomes compared with sodium chloride in patients with chronic kidney disease undergoing an emergent coronary procedure.. Circulation journal : official journal of the Japanese Circulation Society 72(10): 1610–1614 [PubMed: 18756037]
Merten, Gregory J., Burgess, W. Patrick, Gray, Lee V. et al. (2004) Prevention of Contrast-Induced Nephropathy With Sodium BicarbonateA Randomized Controlled Trial. JAMA 291(19): 2328–2334 [PubMed: 15150204]
Miner, Steven E.S., Dzavik, Vladimir, Nguyen-Ho, Phong et al. (2004) N-acetylcysteine reduces contrast-associated nephropathy but not clinical events during long-term follow-up. American Heart Journal 148(4): 690–695 [PubMed: 15459602]
Motohiro M, Kamihata H, Tsujimoto S et al. (2011) A new protocol using sodium bicarbonate for the prevention of contrast-induced nephropathy in patients undergoing coronary angiography.. The American journal of cardiology 107(11): 1604–1608 [PubMed: 21420053]
Mueller, Christian, Buerkle, Gerd, Buettner, Heinz J. et al. (2002) Prevention of contrast media-associated nephropathy: randomized comparison of 2 hydration regimens in 1620 patients undergoing coronary angioplasty. Archives of internal medicine 162(3): 329–36 [PubMed: 11822926]
Nieto-Rios, John Fredy, Salazar, Wilmar Arley Maya, Sanchez, Oscar Mauricio Santos et al. (2014) Prevention of contrast induced nephropathy with sodium bicarbonate (the PROMEC study). Jornal brasileiro de nefrologia : ‘orgao oficial de Sociedades Brasileira e Latino-Americana de Nefrologia 36(3): 360–6 [PubMed: 25317619]
Nijssen, Estelle C., Rennenberg, Roger J., Nelemans, Patty J. et al. (2017) Prophylactic hydration to protect renal function from intravascular iodinated contrast material in patients at high risk of contrast-induced nephropathy (AMACING): a prospective, randomised, phase 3, controlled, open-label, non-inferiority trial. Lancet (London, England) 389(10076): 1312–1322 [PubMed: 28233565]
Oldemeyer, J.Bradley, Biddle, W.Paul, Wurdeman, Richard L et al. (2003) Acetylcysteine in the prevention of contrast-induced nephropathy after coronary angiography. American Heart Journal 146(6): 1089–1094 [PubMed: 14661012]
Poletti PA, Saudan P, Platon A et al. (2007) I.v. N-acetylcysteine and emergency CT: use of serum creatinine and cystatin C as markers of radiocontrast nephrotoxicity.. AJR. American journal of roentgenology 189(3): 687–692 [PubMed: 17715118]
Rashid ST, Salman M, Myint F et al. (2004) Prevention of contrast-induced nephropathy in vascular patients undergoing angiography: a randomized controlled trial of intravenous N-acetylcysteine.. Journal of vascular surgery 40(6): 1136–1141 [PubMed: 15622367]
Reinecke H, Fobker M, Wellmann J et al. (2007) A randomized controlled trial comparing hydration therapy to additional hemodialysis or N-acetylcysteine for the prevention of contrast medium-induced nephropathy: the Dialysis-versus-Diuresis (DVD) Trial.. Clinical research in cardiology : official journal of the German Cardiac Society 96(3): 130–139 [PubMed: 17180572]
Sadineni, R., Karthik, K. R., Swarnalatha, G. et al. (2017) N-acetyl cysteine versus allopurinol in the prevention of contrast nephropathy in patients with chronic kidney disease: A randomized controlled trial. Indian journal of nephrology 27(2): 93–98 [PMC free article: PMC5358166] [PubMed: 28356658]
Saitoh T, Satoh H, Nobuhara M et al. (2011) Intravenous glutathione prevents renal oxidative stress after coronary angiography more effectively than oral N-acetylcysteine.. Heart and vessels 26(5): 465–472 [PubMed: 21127883]
Seyon RA, Jensen LA, Ferguson IA et al. (2007) Efficacy of N-acetylcysteine and hydration versus placebo and hydration in decreasing contrast-induced renal dysfunction in patients undergoing coronary angiography with or without concomitant percutaneous coronary intervention.. Heart & lung : the journal of critical care 36(3): 195–204 [PubMed: 17509426]
Shyu KG; Cheng JJ; Kuan P (2002) Acetylcysteine protects against acute renal damage in patients with abnormal renal function undergoing a coronary procedure.. Journal of the American College of Cardiology 40(8): 1383–1388 [PubMed: 12392825]
Solomon, Richard, Gordon, Paul, Manoukian, Steven V. et al. (2015) Randomized Trial of Bicarbonate or Saline Study for the Prevention of Contrast-Induced Nephropathy in Patients with CKD. Clinical journal of the American Society of Nephrology : CJASN 10(9): 1519–24 [PMC free article: PMC4559510] [PubMed: 26185263]
Tamura, Akira, Goto, Yukie, Miyamoto, Kumie et al. (2009) Efficacy of single-bolus administration of sodium bicarbonate to prevent contrast-induced nephropathy in patients with mild renal insufficiency undergoing an elective coronary procedure. The American journal of cardiology 104(7): 921–5 [PubMed: 19766757]
Tepel M, van der Giet M, Schwarzfeld C et al. (2000) Prevention of radiographic-contrast-agent-induced reductions in renal function by acetylcysteine.. The New England journal of medicine 343(3): 180–184 [PubMed: 10900277]
Thiele H, Hildebrand L, Schirdewahn C et al. (2010) Impact of high-dose N-acetylcysteine versus placebo on contrast-induced nephropathy and myocardial reperfusion injury in unselected patients with ST-segment elevation myocardial infarction undergoing primary percutaneous coronary intervention. The LIPSIA-N-ACC (Prospective, Single-Blind, Placebo-Controlled, Randomized Leipzig Immediate PercutaneouS Coronary Intervention Acute Myocardial Infarction N-ACC) Trial.. Journal of the American College of Cardiology 55(20): 2201–2209 [PubMed: 20466200]
Torigoe, Kumie, Tamura, Akira, Watanabe, Toru et al. (2013) 20-Hour preprocedural hydration is not superior to 5-hour preprocedural hydration in the prevention of contrast-induced increases in serum creatinine and cystatin C. International journal of cardiology 167(5): 2200–3 [PubMed: 22717305]
Traub, Stephen J., Mitchell, Alice M., Jones, Alan E. et al. (2013) N-acetylcysteine plus intravenous fluids versus intravenous fluids alone to prevent contrast-induced nephropathy in emergency computed tomography. Annals of emergency medicine 62(5): 511–520.e25 [PubMed: 23769807]
Turedi, Suleyman, Erdem, Erkan, Karaca, Yunus et al. (2016) The High Risk of Contrast-induced Nephropathy in Patients with Suspected Pulmonary Embolism Despite Three Different Prophylaxis: A Randomized Controlled Trial. Academic emergency medicine : official journal of the Society for Academic Emergency Medicine 23(10): 1136–1145 [PubMed: 27411777]
Ueda, Hiromichi, Yamada, Takahisa, Masuda, Masaharu et al. (2011) Prevention of Contrast-Induced Nephropathy by Bolus Injection of Sodium Bicarbonate in Patients with Chronic Kidney Disease Undergoing Emergent Coronary Procedures. The American Journal of Cardiology 107(8): 1163–1167 [PubMed: 21349483]
van Mourik, M. S., van Kesteren, F., Planken, R. N. et al. (2018) Short versus conventional hydration for prevention of kidney injury during pre-TAVI computed tomography angiography. Netherlands heart journal : monthly journal of the Netherlands Society of Cardiology and the Netherlands Heart Foundation 26(9): 425–432 [PMC free article: PMC6115307] [PubMed: 30039383]
Vasheghani-Farahani A, Sadigh G, Kassaian SE et al. (2010) Sodium bicarbonate in preventing contrast nephropathy in patients at risk for volume overload: a randomized controlled trial.. Journal of nephrology 23(2): 216–223 [PubMed: 20175053]
Webb JG, Pate GE, Humphries KH et al. (2004) A randomized controlled trial of intravenous N-acetylcysteine for the prevention of contrast-induced nephropathy after cardiac catheterization: lack of effect.. American heart journal 148(3): 422–429 [PubMed: 15389228]
Weisbord, Steven D., Gallagher, Martin, Jneid, Hani et al. (2018) Outcomes after Angiography with Sodium Bicarbonate and Acetylcysteine. The New England journal of medicine 378(7): 603–614 [PubMed: 29130810]
Wrobel, Wojciech, Sinkiewicz, Wladyslaw, Gordon, Marcin et al. (2010) Oral versus intravenous hydration and renal function in diabetic patients undergoing percutaneous coronary interventions. Kardiologia polska 68(9): 1015–20 [PubMed: 20859892]

Excluded studies

(2007) MEENA (A Randomized Controlled Trial for the Prevention of Contrast-induced Nephropathy with Sodium Bicarbonate in Persons Undergoing Coronary Angiography). Clinical Cardiology 30(8): 416–416
Agarwal, Shiv Kumar, Mohareb, Sameh, Patel, Achint et al. (2015) Systematic oral hydration with water is similar to parenteral hydration for prevention of contrast-induced nephropathy: an updated meta-analysis of randomised clinical data. Open heart 2(1): e000317 [PMC free article: PMC4600249] [PubMed: 26468404]
Ahmed, K., McVeigh, T., Cerneviciute, R. et al. (2018) Effectiveness of contrast-associated acute kidney injury prevention methods; A systematic review and network meta-analysis. BMC Nephrology 19(1): 323 [PMC free article: PMC6234687] [PubMed: 30424723]
Alessandri, N., Lanzi, L., Garante, C. M. et al. (2013) Prevention of acute renal failure post-contrast imaging in cardiology: a randomized study. European review for medical and pharmacological sciences 17suppl1: 13–21 [PubMed: 23436661]
Ali-Hasan-Al-Saegh, Sadeq, Mirhosseini, Seyed Jalil, Ghodratipour, Zahra et al. (2017) Strategies Preventing Contrast-Induced Nephropathy After Coronary Angiography: A Comprehensive Meta-Analysis and Systematic Review of 125 Randomized Controlled Trials. Angiology 68(5): 389–413 [PubMed: 27485363]
Ali-Hassan-Sayegh, S., Mirhosseini, S. J., Rahimizadeh, E. et al. (2015) Current status of sodium bicarbonate in coronary angiography: An updated comprehensive meta-analysis and systematic review. Cardiology Research and Practice 2015: 690308 [PMC free article: PMC4417980] [PubMed: 25973282]
Alonso, A., Lau, J., Jaber, B. L. et al. (2004) Prevention of Radiocontrast Nephropathy with N-Acetylcysteine in Patients with Chronic Kidney Disease: a Meta-Analysis of Randomized, Controlled Trials. American journal of kidney diseases 43(1): 1–9 [PubMed: 14712421]
Alonso, Pau, Sanz, Jorge, Garcia-Orts, Ana et al. (2017) Usefulness of Sodium Bicarbonate for the Prevention of Contrast-Induced Nephropathy in Patients Undergoing Cardiac Resynchronization Therapy. The American journal of cardiology 120(9): 1584–1588 [PubMed: 28844518]
Bagshaw, S. M. and Ghali, W. A. (2004) Acetylcysteine for prevention of contrast-induced nephropathy after intravascular angiography: a systematic review and meta-analysis. BMC medicine 2: 38 [PMC free article: PMC526263] [PubMed: 15500690]
Bailey, Michael, McGuinness, Shay, Haase, Michael et al. (2015) Sodium bicarbonate and renal function after cardiac surgery: a prospectively planned individual patient meta-analysis. Anesthesiology 122(2): 294–306 [PubMed: 25501691]
Balderramo D, Verdu M, Ramacciotti. C.F. et al. (2004) Renoprotective 30 effect of high periprocedural doses of oral N-acetylcysteine in patients scheduled to undergo a same-day angiography. Revista De La Facultad De Ciencias Medicas De Cordoba 61(2) [PubMed: 16211988]
Berwanger, Otavio, Cavalcanti, Alexandre Biasi, Sousa, Amanda M. G. et al. (2013) Acetylcysteine for the prevention of renal outcomes in patients with diabetes mellitus undergoing coronary and peripheral vascular angiography: a substudy of the acetylcysteine for contrast-induced nephropathy trial. Circulation. Cardiovascular interventions 6(2): 139–45 [PubMed: 23572490]
Biernacka-Fialkowska, Barbara, Szuksztul, Marta, Suslik, Wojciech et al. (2018) Intravenous N-acetylcysteine for the PRevention Of Contrast-induced nephropathy - a prospective, single-center, randomized, placebo-controlled trial. The INPROC trial. Postepy w kardiologii interwencyjnej = Advances in interventional cardiology 14(1): 59–66 [PMC free article: PMC5939546] [PubMed: 29743905]
Boccalandro, F., Amhad, M., Smalling, R. W. et al. (2003) Oral acetylcysteine does not protect renal function from moderate to high doses of intravenous radiographic contrast. Catheterization and cardiovascular interventions 58(3): 336–341 [PubMed: 12594698]
Brown, Jeremiah R., Pearlman, Daniel M., Marshall, Emily J. et al. (2016) Meta-Analysis of Individual Patient Data of Sodium Bicarbonate and Sodium Chloride for All-Cause Mortality After Coronary Angiography. The American journal of cardiology 118(10): 1473–1479 [PMC free article: PMC6579735] [PubMed: 27642111]
Brueck, Martin, Cengiz, Huelya, Hoeltgen, Reinhard et al. (2013) Usefulness of N-acetylcysteine or ascorbic acid versus placebo to prevent contrast-induced acute kidney injury in patients undergoing elective cardiac catheterization: a single-center, prospective, randomized, double-blind, placebo-controlled trial. The Journal of invasive cardiology 25(6): 276–83 [PubMed: 23735352]
Burns KE; Priestap F; Martin C (2010) N-acetylcysteine in critically ill patients undergoing contrast-enhanced computed tomography: a randomized trial.. Clinical nephrology 74(4): 323–326 [PubMed: 20875388]
Busch, Sarah Victoria Ekelof, Jensen, Svend Eggert, Rosenberg, Jacob et al. (2013) Prevention of contrast-induced nephropathy in STEMI patients undergoing primary percutaneous coronary intervention: a systematic review. Journal of interventional cardiology 26(1): 97–105 [PubMed: 23240788]
Chen, H. W., Zhang, J. J., Xiong, D. et al. (2016) Prevention and Treatment of Shenkang Injection for Contrast-induced Nephropathy in Elder Patients with Chronic Kidney Disease. Zhongguo zhong xi yi jie he za zhi zhongguo zhongxiyi jiehe zazhi = chinese journal of integrated traditional and western medicine 36(7): 792–796 [PubMed: 30634203]
Cheungpasitporn, Wisit, Thongprayoon, Charat, Brabec, Brady A. et al. (2014) Oral hydration for prevention of contrast-induced acute kidney injury in elective radiological procedures: a systematic review and meta-analysis of randomized controlled trials. North American journal of medical sciences 6(12): 618–24 [PMC free article: PMC4290050] [PubMed: 25599049]
Dabare, Dilan, Banihani, Mohammed, Gibbs, Paul et al. (2013) Does bicarbonate prevent contrast-induced nephropathy in cardiovascular patients undergoing contrast imaging?. Interactive cardiovascular and thoracic surgery 17(6): 1028–35 [PMC free article: PMC3829505] [PubMed: 23996732]
Ding, L.; Zhuang, G. H.; Ding, B. (2016) Clinical application of intravenous hydration or oral hydration in preventing contrast-induced nephropathy in patients with cardiac insufficiency. Journal of interventional radiology (china) 25(1): 15–18
Dong, Yuhao, Zhang, Bin, Liang, Long et al. (2016) How Strong Is the Evidence for Sodium Bicarbonate to Prevent Contrast-Induced Acute Kidney Injury After Coronary Angiography and Percutaneous Coronary Intervention?. Medicine 95(7): e2715 [PMC free article: PMC4998610] [PubMed: 26886610]
Eskandarian, R., Yarmohamadi, M., Zaker-Tavalae, M. et al. (2018) The standard dose versus double dose of n-acetylcysteine to prevent contrast-induced nephropathy; a randomized controlled clinical trial. Journal of Nephropathology 7(3): 145–150
Giacoppo, Daniele, Gargiulo, Giuseppe, Buccheri, Sergio et al. (2017) Preventive Strategies for Contrast-Induced Acute Kidney Injury in Patients Undergoing Percutaneous Coronary Procedures: Evidence From a Hierarchical Bayesian Network Meta-Analysis of 124 Trials and 28 240 Patients. Circulation. Cardiovascular interventions 10(5) [PubMed: 28487354]
Glaza, M.; Rutkowski, B.; Szolkiewicz, M. (2018) Prevention of contrast-induced nephropathy in patients after percutaneous coronary intervention: A single-center prospective study. Clinical Nephrology 90(5): 370–372 [PubMed: 30049301]
Guru, V. and Fremes, S. E. (2004) The role of N-acetylcysteine in preventing radiographic contrast-induced nephropathy. Clinical nephrology 62(2): 77–83 [PubMed: 15356963]
Heguilen, Ricardo M., Liste, Amador A., Payaslian, Miguel et al. (2013) N-acethyl-cysteine reduces the occurrence of contrast-induced acute kidney injury in patients with renal dysfunction: a single-center randomized controlled trial. Clinical and experimental nephrology 17(3): 396–404 [PubMed: 23138396]
Hiremath, Swapnil, Akbari, Ayub, Shabana, Wael et al. (2013) Prevention of contrast-induced acute kidney injury: is simple oral hydration similar to intravenous? A systematic review of the evidence. PloS one 8(3): e60009 [PMC free article: PMC3608617] [PubMed: 23555863]
Inda-Filho, Antonio Jose, Caixeta, Adriano, Manggini, Marcia et al. (2014) Do intravenous N-acetylcysteine and sodium bicarbonate prevent high osmolal contrast-induced acute kidney injury? A randomized controlled trial. PloS one 9(9): e107602 [PMC free article: PMC4177831] [PubMed: 25254489]
Izcovich, Ariel and Rada, Gabriel (2015) Should acetylcysteine be used to prevent contrast induced nephropathy?. Medwave 15(3): e6122 [PubMed: 25918998]
Jang, Jae-Sik, Jin, Han-Young, Seo, Jeong-Sook et al. (2012) Sodium bicarbonate therapy for the prevention of contrast-induced acute kidney injury - a systematic review and meta-analysis. Circulation journal : official journal of the Japanese Circulation Society 76(9): 2255–65 [PubMed: 22975638]
Jiang, Yufeng, Chen, Min, Zhang, Yiqing et al. (2017) Meta-analysis of prophylactic hydration versus no hydration on contrast-induced acute kidney injury. Coronary artery disease 28(8): 649–657 [PubMed: 28692484]
Jurado-Roman, Alfonso, Hernandez-Hernandez, Felipe, Garcia-Tejada, Julio et al. (2015) Role of hydration in contrast-induced nephropathy in patients who underwent primary percutaneous coronary intervention. The American journal of cardiology 115(9): 1174–8 [PubMed: 25759106]
Kanbay, M., Covic, A., Coca, S. G. et al. (2009) Sodium bicarbonate for the prevention of contrast-induced nephropathy: a meta-analysis of 17 randomized trials. International urology and nephrology 41(3): 617–627 [PubMed: 19396567]
Kang, Xin, Hu, Da-Yong, Li, Chang-Bin et al. (2015) N-acetylcysteine for the prevention of contrast-induced nephropathy in patients with pre-existing renal insufficiency or diabetes: a systematic review and meta-analysis. Renal failure 37(10): 297–303 [PubMed: 26458505]
Khaledifar, A., Momeni, A., Ebrahimi, A. et al. (2015) Comparison of N-acetylcysteine, ascorbic acid, and normal saline effect in prevention of contrast-induced nephropathy. ARYA Atherosclerosis 11(4) [PMC free article: PMC4593658] [PubMed: 26478730]
Khan, Safi U., Khan, Muhammad U., Rahman, Hammad et al. (2019) A Bayesian network meta-analysis of preventive strategies for contrast-induced nephropathy after cardiac catheterization. Cardiovascular revascularization medicine : including molecular interventions 20(1): 29–37 [PubMed: 30757995]
Kim, Byung Jin, Sung, Ki Chul, Kim, Bum Soo et al. (2010) Effect of N-Acetylcysteine on cystatin C-Based renaL function after Elective coronary angiography (ENABLE Study): A prospective, randomized trial. International Journal of Cardiology 138(3): 239–245 [PubMed: 18793808]
Kimmel, Martin, Butscheid, Moritz, Brenner, Stefanie et al. (2008) Improved estimation of glomerular filtration rate by serum cystatin C in preventing contrast induced nephropathy by N-acetylcysteine or zinc?preliminary results. ndt 23(4): 1241–1245 [PubMed: 18174269]
Koc, Fatih, Ozdemir, Kurtulus, Altunkas, Fatih et al. (2013) Sodium bicarbonate versus isotonic saline for the prevention of contrast-induced nephropathy in patients with diabetes mellitus undergoing coronary angiography and/or intervention: a multicenter prospective randomized study. Journal of investigative medicine : the official publication of the American Federation for Clinical Research 61(5): 872–7 [PubMed: 23552179]
Kumar, A., Bhawani, G., Kumari, N. et al. (2014) Comparative study of renal protective effects of allopurinol and n-acetyl-cysteine on contrast induced nephropathy in patients undergoing cardiac catheterization. Journal of Clinical and Diagnostic Research 8(12): HC03–HC07 [PMC free article: PMC4316271] [PubMed: 25653965]
Li, J., Jin, E., Yu, L. et al. (2017) Oral N-acetylcysteine for prophylaxis of contrast-induced nephropathy in patients following coronary angioplasty: A meta-analysis. Experimental and Therapeutic Medicine 14(2): 1568–1576 [PMC free article: PMC5525578] [PubMed: 28810622]
Liu, R., Nair, D., Ix, J. et al. (2005) N-acetylcysteine for the prevention of contrast-induced nephropathy: a systematic review and meta-analysis. Journal of General Internal Medicine 20(2): 193–200 [PMC free article: PMC1490056] [PubMed: 15836554]
Loomba, R. S., Shah, P. H., Aggarwal, S. et al. (2013) Role of N-acetylcysteine to prevent contrast-induced nephropathy: a meta-analysis. American Journal of Therapeutics: epub [PubMed: 23982694]
Loomba, Rohit S., Shah, Parinda H., Aggarwal, Saurabh et al. (2016) Role of N-Acetylcysteine to Prevent Contrast-Induced Nephropathy: A Meta-analysis. American journal of therapeutics 23(1): e172–83 [PubMed: 23982694]
Luo, Yu, Wang, Xiaodong, Ye, Zi et al. (2014) Remedial hydration reduces the incidence of contrast-induced nephropathy and short-term adverse events in patients with ST-segment elevation myocardial infarction: a single-center, randomized trial. Internal medicine (Tokyo, Japan) 53(20): 2265–72 [PubMed: 25318787]
Ma, W. Q., Zhao, Y., Wang, Y. et al. (2018) Comparative efficacy of pharmacological interventions for contrast-induced nephropathy prevention after coronary angiography: a network meta-analysis from randomized trials. International Urology and Nephrology 50(6): 1085–1095 [PubMed: 29404930]
Mahmoodi, Khalil, Sohrabi, Bahram, Ilkhchooyi, Farzad et al. (2014) The Efficacy of Hydration with Normal Saline Versus Hydration with Sodium Bicarbonate in the Prevention of Contrast-induced Nephropathy. Heart views : the official journal of the Gulf Heart Association 15(2): 33–6 [PMC free article: PMC4124663] [PubMed: 25104980]
Manari, Antonio, Magnavacchi, Paolo, Puggioni, Enrico et al. (2014) Acute kidney injury after primary angioplasty: effect of different hydration treatments. Journal of cardiovascular medicine (Hagerstown, Md.) 15(1): 60–7 [PubMed: 24500238]
Navarese, Eliano P., Gurbel, Paul A., Andreotti, Felicita et al. (2017) Prevention of contrast-induced acute kidney injury in patients undergoing cardiovascular procedures-a systematic review and network meta-analysis. PloS one 12(2): e0168726 [PMC free article: PMC5289438] [PubMed: 28151965]
O’Sullivan, S., Healy, D. A., Moloney, Mary Clarke et al. (2013) The role of N--acetylcysteine in the prevention of contrast-induced nephropathy in patients undergoing peripheral angiography: a structured review and meta-analysis. Angiology 64(8): 576–82 [PubMed: 23188834]
Pakfetrat M, Nikoo MH, Malekmakan L et al. (2009) A comparison of sodium bicarbonate infusion versus normal saline infusion and its combination with oral acetazolamide for prevention of contrast-induced nephropathy: a randomized, double-blind trial.. International urology and nephrology 41(3): 629–634 [PubMed: 19137409]
Pakfetrat, Maryam, Malekmakan, Leila, Salmanpour, Zahra et al. (2019) Comparison of Normal Saline, Ringer’s Lactate, and Sodium Bicarbonate for Prevention of Contrast-induced Nephropathy in Patients with Coronary Angiography: A Randomized Double-blind Clinical Trial. Indian journal of nephrology 29(1): 22–27 [PMC free article: PMC6375023] [PubMed: 30814789]
Pandya, B., Chaloub, J., Parikh, V. et al. (2017) Contrast media use in patients with chronic kidney disease undergoing coronary angiography: A systematic review and meta-analysis of randomized trials. International Journal of Cardiology 228: 137–144 [PubMed: 27863354]
Pezeshgi, Aiyoub, Parsamanesh, Negin, Farhood, Goodarz et al. (2015) Evaluation of the protective effect of N-acetylcysteine on contrast media nephropathy. Journal of renal injury prevention 4(4): 109–12 [PMC free article: PMC4685979] [PubMed: 26693496]
Poletti, Pierre-Alexandre, Platon, Alexandra, De Seigneux, Sophie et al. (2013) N-acetylcysteine does not prevent contrast nephropathy in patients with renal impairment undergoing emergency CT: a randomized study. BMC nephrology 14: 119 [PMC free article: PMC3682900] [PubMed: 23731573]
Ratcliffe JA, Thiagarajah P, Chen J et al. (2009) Prevention of contrast-induced nephropathy: A randomized controlled trial of sodium bicarbonate and N-acetylcysteine.. The International journal of angiology : official publication of the International College of Angiology, Inc 18(4): 193–197 [PMC free article: PMC2903033] [PubMed: 22477552]
Sadat U, Walsh SR, Norden AG et al. (2011) Does oral N-acetylcysteine reduce contrast-induced renal injury in patients with peripheral arterial disease undergoing peripheral angiography? A randomized-controlled study.. Angiology 62(3): 225–230 [PubMed: 20682612]
Sar F, Saler T, Ecebay A et al. (2010) The efficacy of n-acetylcysteine in preventing contrast-induced nephropathy in type 2 diabetic patients without nephropathy.. Journal of nephrology 23(4): 478–482 [PubMed: 20383874]
Sharp, Alexander J., Patel, Nishith, Reeves, Barney C. et al. (2019) Pharmacological interventions for the prevention of contrast-induced acute kidney injury in high-risk adult patients undergoing coronary angiography: a systematic review and meta-analysis of randomised controlled trials. Open heart 6(1): e000864 [PMC free article: PMC6350720] [PubMed: 30774964]
Silva RG, Silva NG, Lucchesi F et al. (2010) Prevention of contrast-induced nephropathy by use of bicarbonate solution: preliminary results and literature review.. Jornal brasileiro de nefrologia : ‘orgao oficial de Sociedades Brasileira e Latino-Americana de Nefrologia 32(3): 292–302 [PubMed: 21103694]
Sinert, R. and Doty, C. I. (2009) Update: prevention of contrast-induced nephropathy in the emergency department. Annals of Emergency Medicine 54(1): e1–e5 [PubMed: 18926598]
Su, Xiaole, Xie, Xinfang, Liu, Lijun et al. (2017) Comparative Effectiveness of 12 Treatment Strategies for Preventing Contrast-Induced Acute Kidney Injury: A Systematic Review and Bayesian Network Meta-analysis. American journal of kidney diseases : the official journal of the National Kidney Foundation 69(1): 69–77 [PubMed: 27707552]
Subramaniam, Rathan M., Suarez-Cuervo, Catalina, Wilson, Renee F. et al. (2016) Effectiveness of Prevention Strategies for Contrast-Induced Nephropathy: A Systematic Review and Meta-analysis. Annals of internal medicine 164(6): 406–16 [PubMed: 26830221]
Subramaniam, Rathan M., Wilson, Renee F., Turban, Sharon et al. (2016) Contrast-Induced Nephropathy: Comparative Effectiveness of Preventive Measures. Agency for Healthcare Research and Quality (US). AHRQ Comparative Effectiveness Reviews, Report No.: 15(16)-EHC023-EF [PubMed: 26866209]
Sun, Zikai, Fu, Qiang, Cao, Longxing et al. (2013) Intravenous N-acetylcysteine for prevention of contrast-induced nephropathy: a meta-analysis of randomized, controlled trials. PloS one 8(1): e55124 [PMC free article: PMC3559541] [PubMed: 23383076]
Tanaka A, Suzuki Y, Suzuki N et al. (2011) Does N-acetylcysteine reduce the incidence of contrast-induced nephropathy and clinical events in patients undergoing primary angioplasty for acute myocardial infarction?. Internal medicine (Tokyo, Japan) 50(7): 673–677 [PubMed: 21467697]
Thayssen, Per, Lassen, Jens Flensted, Jensen, Svend Eggert et al. (2014) Prevention of contrast-induced nephropathy with N-acetylcysteine or sodium bicarbonate in patients with ST-segment-myocardial infarction: a prospective, randomized, open-labeled trial. Circulation. Cardiovascular interventions 7(2): 216–24 [PubMed: 24714489]
Trivedi HS, Moore H, Nasr S et al. (2003) A randomized prospective trial to assess the role of saline hydration on the development of contrast nephrotoxicity.. Nephron. Clinical practice 93(1): c29 [PubMed: 12411756]
Valette, Xavier, Desmeulles, Isabelle, Savary, Benoit et al. (2017) Sodium Bicarbonate Versus Sodium Chloride for Preventing Contrast-Associated Acute Kidney Injury in Critically Ill Patients: A Randomized Controlled Trial. Critical care medicine 45(4): 637–644 [PubMed: 28181941]
Wang, Nelson, Qian, Pierre, Kumar, Shejil et al. (2016) The effect of N-acetylcysteine on the incidence of contrast-induced kidney injury: A systematic review and trial sequential analysis. International journal of cardiology 209: 319–27 [PubMed: 26922293]
Weisbord, Steven D., Gallagher, Martin, Kaufman, James et al. (2013) Prevention of contrast-induced AKI: a review of published trials and the design of the prevention of serious adverse events following angiography (PRESERVE) trial. Clinical journal of the American Society of Nephrology : CJASN 8(9): 1618–31 [PMC free article: PMC3805082] [PubMed: 23660180]
Wu, Mei-Yi, Hsiang, Hui-Fen, Wong, Chung-Shun et al. (2013) The effectiveness of N-Acetylcysteine in preventing contrast-induced nephropathy in patients undergoing contrast-enhanced computed tomography: a meta-analysis of randomized controlled trials. International urology and nephrology 45(5): 1309–18 [PubMed: 23283594]
Xu, Renfan, Tao, Anyu, Bai, Yang et al. (2016) Effectiveness of N-Acetylcysteine for the Prevention of Contrast-Induced Nephropathy: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Journal of the American Heart Association 5(9) [PMC free article: PMC5079043] [PubMed: 27663415]
Yang, Kun, Liu, Wenxian, Ren, Wei et al. (2014) Different interventions in preventing contrast-induced nephropathy after percutaneous coronary intervention. International urology and nephrology 46(9): 1801–7 [PubMed: 24966097]
Yeganehkhah, Mohammad Reza, Iranirad, Leili, Dorri, Farshad et al. (2014) Comparison between three supportive treatments for prevention of contrast-induced nephropathy in high-risk patients undergoing coronary angiography. Saudi journal of kidney diseases and transplantation : an official publication of the Saudi Center for Organ Transplantation, Saudi Arabia 25(6): 1217–23 [PubMed: 25394438]
Zagler, Axel, Azadpour, Maziar, Mercado, Carlos et al. (2006) N-acetylcysteine and contrast-induced nephropathy: a meta-analysis of 13 randomized trials. American heart journal 151(1): 140–5 [PubMed: 16368307]
Zapata-Chica, Carlos Andres, Bello Marquez, Diana, Serna-Higuita, Lina Maria et al. (2015) Sodium bicarbonate versus isotonic saline solution to prevent contrast-induced nephropathy : a systematic review and meta-analysis. Colombia medica (Cali, Colombia) 46(3): 90–103 [PMC free article: PMC4640430] [PubMed: 26600623]
Zhang, Bin, Liang, Long, Chen, Wenbo et al. (2015) The efficacy of sodium bicarbonate in preventing contrast-induced nephropathy in patients with pre-existing renal insufficiency: a meta-analysis. BMJ open 5(3): e006989 [PMC free article: PMC4368906] [PubMed: 25783425]
Zhao, Shi-Jie, Zhong, Zhao-Shuang, Qi, Guo-Xian et al. (2016) The efficacy of N-acetylcysteine plus sodium bicarbonate in the prevention of contrast-induced nephropathy after cardiac catheterization and percutaneous coronary intervention: A meta-analysis of randomized controlled trials. International journal of cardiology 221: 251–9 [PubMed: 27404685]
Zoungas, S., Ninomiya, T., Huxley, R. et al. (2009) Systematic review: sodium bicarbonate treatment regimens for the prevention of contrast-induced nephropathy. Annals of internal medicine 151(9): 631–638 [PubMed: 19884624]

Economic studies

Included studies

None

Excluded studies

Kooiman, de Vries, Van der Heyden, Sijpkens, van Dijkman, Wever et al. (2018) Randomized trial of one-hour sodium bicarbonate vs standard periprocedural saline hydration in chronic kidney disease patients undergoing cardiovascular contrast procedures. PloS one 13(2): e0189372. [PMC free article: PMC5805164] [PubMed: 29420536]
Kooiman, Sijpkens, de Vries, Brulez, Hamming, van der Molen et al. (2014) A randomized comparison of 1-h sodium bicarbonate hydration versus standard peri-procedural saline hydration in patients with chronic kidney disease undergoing intravenous contrast-enhanced computerized tomography. European Renal Association 29(5); 1029–36. [PubMed: 24578471]
Kotlyar, Keogh, Thavapalachandran, Allada, Sharp, Dias et al. (2005) Prehydration alone is sufficient to prevent contrast-induced nephropathy after day-only angiography procedures--a randomised controlled trial. Heart, lung & circulation 14(4); 245–51. [PubMed: 16360994]
Nijssen, Rennenberg, Nelemans, Essers, Janssen, Vermeeren et al. (2017) Prophylactic hydration to protect renal function from intravascular iodinated contrast material in patients at high risk of contrast-induced nephropathy (AMACING): a prospective, randomised, phase 3, controlled, open-label, non-inferiority trial. Lancet 389: 1312–22 [PubMed: 28233565]

Appendix P. List of CI-AKI definitions

Table 34. List of CI-AKI definitions reported by included studies

Appendix Q. NMA models

Please refer to appendix R for the inconsistency models.

Fixed effects model for relative risk with input and output codes swapped

# Key data inputs: absID=9, outID = c(2, 16, 17, 14, 12, 15, 10, 13, 9, 6, 3, 7, 4, 8, 1, 11, 5) 
model{ # *** PROGRAM STARTS 
for(i in 1:ns){ # LOOP THROUGH STUDIES 
  mu[i] ~ dnorm(0,.0001) # vague priors for all trial baselines 
  for (k in 1:na[i]) { # LOOP THROUGH ARMS 62- can do > 2 arms 
    r[i,k] ~ dbin(p[i,k],n[i,k]) # binomial likelihood 
    logit(p[i,k]) <- mu[i] + d[t[i,k]] - d[t[i,1]] # model for linear predictor 
    rhat[i,k] <- p[i,k] * n[i,k] # expected value of the numerators 
    dev[i,k] <- 2 * (r[i,k] * (log(r[i,k])-log(rhat[i,k])) #Deviance contribution 
    + (n[i,k]-r[i,k]) * (log(n[i,k]-r[i,k]) - log(n[i,k]-rhat[i,k]))) 
    } 
  resdev[i] <- sum(dev[i,1:na[i]]) # summed residual deviance contribution for this trial 
  } 
totresdev <- sum(resdev[]) #Total Residual Deviance 
d[1] <- 0 # treatment effect is zero for reference treatment 
for (k in 2:nt){ d[k] ~ dnorm(0,.0001) } # vague priors for treatment effects 
# reorder effects according to vector outID[] 
for (k in 1:nt) { 
  d3[k] <- d[outID[k]] - d[outID[1]] 
  } 
An <- 250 # estimate of absolute AKI prob with NaCl(0.9%)+NAC from Maioli et al. (2008) 
Ak <- 25 
Ab <- An - Ak 
A ~ dbeta(Ak, Ab) 
for (k in 1:nt) { 
  logit(T[k]) <- logit(A) + d3[k] + d[outID[1]] - d[absID] 
  } 
# RR for each treatment relative to reference option (use for caterpillar plots) 
RR[1] <- 1 
for (k in 2:nt) { 
  RR[k] <- T[k]/T[1] 
  } 
# pairwise ORs and RRs 
for (c in 1:(nt-1)) { 
  for (k in (c+1):nt) { 
    OR[c,k] <- exp(d3[k] - d3[c]) 
    RRR[c,k] <- T[k]/T[c] 
    } 
  } 
# rank treatments 
for (k in 1:nt)  { 
  rk[k]  <- rank(d3[],k) 
  best[k]  <- equals(rk[k],1)    # Smallest is best (i.e. rank 1) 
  for (h in 1:nt) { 
    prob[h,k] <- equals(rk[k],h) 
    } 
  } 
} # *** PROGRAM ENDS

Random effects model for relative risk with input and output codes swapped

# Key data inputs: absID=9, outID = c(2, 16, 17, 14, 12, 15, 10, 13, 9, 6, 3, 7, 4, 8, 1, 11, 5) 
model{ # *** PROGRAM STARTS 
for(i in 1:ns){ # LOOP THROUGH STUDIES 
  w[i,1] <- 0 # adjustment for multi-arm trials is zero for control arm 
  delta[i,1] <- 0 # treatment effect is zero for control arm 
  mu[i] ~ dnorm(0,.0001) # vague priors for all trial baselines 
  for (k in 1:na[i]) { # LOOP THROUGH ARMS 
    r[i,k] ~ dbin(p[i,k],n[i,k]) # binomial likelihood 
    logit(p[i,k]) <- mu[i] + delta[i,k] # model for linear predictor 
    rhat[i,k] <- p[i,k] * n[i,k] # expected value of the numerators 
    dev[i,k] <- 2 * (r[i,k] * (log(r[i,k])-log(rhat[i,k])) #Deviance contribution 
    + (n[i,k]-r[i,k]) * (log(n[i,k]-r[i,k]) - log(n[i,k]-rhat[i,k]))) 
    } 
  resdev[i] <- sum(dev[i,1:na[i]]) # summed residual deviance contribution for this trial 
  for (k in 2:na[i]) { # LOOP THROUGH ARMS 
    delta[i,k] ~ dnorm(md[i,k],taud[i,k]) # trial-specific LOR distributions 
    md[i,k] <- d[t[i,k]] - d[t[i,1]] + sw[i,k] # mean of LOR distributions (with multi-arm trial correction) 
    w[i,k] <- (delta[i,k] - d[t[i,k]] + d[t[i,1]]) # adjustment for multi-arm RCTs 
    taud[i,k] <- tau *2*(k-1)/k # precision of LOR distributions (with multi-arm trial correction) 
    sw[i,k] <- sum(w[i,1:k-1])/(k-1) # cumulative adjustment for multi-arm trials 
    } 
  } 
totresdev <- sum(resdev[]) #Total Residual Deviance 
sd ~ dunif(0,5) # vague prior for between-trial SD. ALTERNATIVES BELOW 
tau <- pow(sd,-2) # between-trial precision = (1/between-trial variance) 
An <- 250 # estimate of absolute AKI prob with NaCl(0.9%)+NAC from Maioli et al. (2008) 
Ak <- 25 
Ab <- An - Ak 
A ~ dbeta(Ak, Ab) 
for (k in 1:nt) { 
  logit(T[k]) <- logit(A) + d3[k] + d[outID[1]] - d[absID] 
  }  
# RR for each treatment relative to reference option (use for caterpillar plots) 
RR[1] <- 1 
for (k in 2:nt) { 
  RR[k] <- T[k]/T[1] 
  } 
# pairwise ORs and LORs for all possible pair-wise comparisons, if nt>2 
for (c in 1:(nt-1)) { 
for (k in (c+1):nt) { 
or[c,k] <- exp(d[k] - d[c]) 
lor[c,k] <- (d[k]-d[c]) 
} 
} 
# pairwise ORs and RRs 
for (c in 1:(nt-1)) { 
  for (k in (c+1):nt) { 
    OR[c,k] <- exp(d3[k] - d3[c]) 
    RRR[c,k] <- T[k]/T[c] 
    } 
  } 
# rank treatments 
for (k in 1:nt)  { 
  rk[k]  <- rank(d3[],k) 
  best[k]  <- equals(rk[k],1)    # Smallest is best (i.e. rank 1) 
  for (h in 1:nt) { 
    prob[h,k] <- equals(rk[k],h) 
    } 
  } 
d[1] <- 0 # treatment effect is zero for reference treatment 
for (k in 2:nt){ d[k] ~ dnorm(0,.0001) } # vague priors for treatment effects 
# reorder effects according to vector outID[] 
for (k in 1:nt) { 
  d3[k] <- d[outID[k]] - d[outID[1]] 
  } 
} # *** PROGRAM ENDS

Appendix R. NMA inconsistency checks

Introduction

The purpose of this analysis was to assess the consistency assumption in the network meta-analysis (NMA) model used to estimate the clinical and cost effectiveness of N-acetylcysteine (NAC) and/or fluids in preventing contrast induced acute kidney injury (CI-AKI) in at risk adults. Ocurrence of CI-AKI was the only outcome included in this analysis (see section ‘NMA analyses and NMA model inconsistency checks’ for more details about the inconsistency checks).

Methods

An important assumption made in NMA concerns the consistency of the direct and indirect evidence informing the treatment contrasts [1,2]. There should be no meaningful differences between these two sources of evidence.

To determine if there is evidence of inconsistency, the selected consistency model (fixed or random effects) was compared to an “inconsistency”, or unrelated mean effects, model [1,2]. The latter is equivalent to having separate, unrelated, meta-analyses for every pairwise contrast, with a common variance parameter assumed in the case of random effects models. Note that the consistency assumption can only be assessed when there are closed loops of direct evidence on 3 treatments that are informed by at least 3 independent sources of evidence [3].

The posterior mean of the residual deviance, which measures the magnitude of the differences between the observed data and the model predictions of the data, was used to assess and compare the goodness of fit of each model [4]. Smaller values are preferred, and in a well-fitting model the posterior mean residual deviance should be close to the number of data points in the network (each study arm contributes 1 data point) [4].

In addition to comparing how well the models fit the data using the posterior mean of the residual deviance, models were compared using the deviance information criterion (DIC). This is equal to the sum of the posterior mean of the residual deviance and the effective number of parameters, and thus penalizes model fit with model complexity [4]. Lower values are preferred and typically differences of 3–5 points are considered meaningful [4].

The posterior mean between-study standard deviation, which measures the heterogeneity of treatment effects estimated by trials within contrasts, was also used to compare models. when comparing consistency and inconsistency models, if the inconsistency model has the smallest heterogeneity, then this indicates potential inconsistency in the data.

We performed further checks for evidence of inconsistency through node-splitting [1–3,5]. This method permits the direct and indirect evidence contributing to an estimate of a relative effect to be split and compared.

Results

Inconsistency checks were performed using the random effects model, as lower posterior mean residual deviance and DIC models compared to the fixed effect model suggest the random effects model provided a better fit for the data (Table 35).

Table 35. Model fit statistics

Since there were closed loops of direct evidence within the network that were informed by at least 3 distinct sets of trials, inconsistency checks were possible for this outcome. Convergence was satisfactory for the random effects model assuming inconsistency after 50,000 iterations, and the consistency and inconsistency models were compared using results based on samples from a further 50,000 iterations on two chains. WinBUGS code for the inconsistency model is provided in Appendix R.1. WinBUGS code for inconsistency model used in this report

No evidence of inconsistency was found through comparison of the consistency and inconsistency random effects models, as little difference was observed between the fit of the models (Table 35). The area below the line of equality in Figure 34 highlights where the inconsistency model better predicted data points, and there were notable improvements in the prediction of data in Ueda 2011 which included 2 of the treatments in the loop of evidence identified as potentially inconsistent in the node-splitting analysis (treatments: 3=sodium bicarbonate [IV] and 11=sodium chloride 0.9% [IV] + sodium bicarbonate [IV]). Ueda 2011 was better predicted by the inconsistency model. There were no errors in data extraction for Ueda 2011. Therefore, Ueda 2011 was removed from the data and the NMA models were run again (consistency and inconsistency models). Results were not too different compared to the models including all RCTs. Therefore, the committee decided to use the results with all RCTs when they discussed the evidence. The additional parameters in the inconsistency model, which eliminates variation between treatment contrasts, did not result in a significant decrease in the between-study heterogeneity (Table 35). See section of NMA analyses and NMA model inconsistency checks for a description of the inconsistency checks.

Figure 34. Deviance contributions for the random effects consistency and inconsistency models

Further checks for inconsistency using the node-splitting method (random effects model) found that there was evidence of inconsistency in the 1-3-11 loop, where 1 = sodium chloride 0.9% (IV), 3=sodium bicarbonate (IV), 11=sodium chloride 0.9% (IV) + sodium bicarbonate (IV). (Table 36, Figure 35–37). All RCTs involved in the 1-3-11 loop were scrutinised (no errors were found in data extraction) and there were no obvious differences in study characteristics that could account for the inconsistency. In addition to the relative effects estimated through NMA, we present direct (when available) and indirect estimates inTable 37. Where direct evidence is available on treatment comparisons, the direct and indirect estimates are reported based on results given by the node-splitting models. Otherwise, the indirect estimates are taken from the NMA model. All NMA estimates are reported based on the results from the random effects model that assumes consistency [6,7].

Table 36. Summary of node-splitting results

Figure 35. Direct, indirect and network estimates of relative treatment effects based on node-splitting results (Part 1)

Figure 36. Direct, indirect and network estimates of relative treatment effects based on node-splitting results (Part 2)

Figure 37. Direct, indirect and network estimates of relative treatment effects based on node-splitting results (Part 3)

Table 37. Direct, indirect and NMA estimates of all relative treatment effects

Conclusion

There was evidence of inconsistency in the network. Data from Ueda 2011 and from the other studies involved in the 1-3-11 loop was scrutinised to ensure there were no errors that could account for this issue, but none were found. The committee expected that there would be inconsistency in the results because they were aware of intervention-level differences but these differences were accepted as part of the wide range of fluids regimens. Further checks for inconsistency using the node-splitting method (random effects model) found evidence of inconsistency between the direct and indirect estimates for two treatment comparisons: 1) sodium chloride 0.9% (IV) vs sodium chloride 0.9% (IV) + sodium bicarbonate (IV) and 2) sodium chloride 0.9% (IV) + sodium bicarbonate (IV) vs sodium bicarbonate (IV). Caution should be exercised when interpreting the results.

Appendix R.1. WinBUGS code for inconsistency model used in this report

# Binomial likelihood, logit link 
# Random effects inconsistency model 
model{ # *** PROGRAM STARTS 
for(i in 1:ns){ # LOOP THROUGH STUDIES 
  w[i,1] <- 0 # adjustment for multi-arm trials is zero for control arm 
  delta[i,1] <- 0 # treatment effect is zero for control arm 
  mu[i] ~ dnorm(0,.0001) # vague priors for all trial baselines 
  for (k in 1:na[i]) { # LOOP THROUGH ARMS 
    r[i,k] ~ dbin(p[i,k],n[i,k]) # binomial likelihood 
    logit(p[i,k]) <- mu[i] + delta[i,k] # model for linear predictor 
    rhat[i,k] <- p[i,k] * n[i,k] # expected value of the numerators 
    dev[i,k] <- 2 * (r[i,k] * (log(r[i,k])-log(rhat[i,k])) #Deviance contribution 
    + (n[i,k]-r[i,k]) * (log(n[i,k]-r[i,k]) - log(n[i,k]-rhat[i,k]))) 
    } 
  resdev[i] <- sum(dev[i,1:na[i]]) # summed residual deviance contribution for this trial 
  for (k in 2:na[i]) { # LOOP THROUGH ARMS 
    delta[i,k] ~ dnorm(d[t[i,1],t[i,k]],tau) # trial-specific LOR distributions 
    } 
  } 
totresdev <- sum(resdev[]) #Total Residual Deviance 
for (c in 1:nt) {   d[c,c] <- 0 } 
for (c in 1:(nt-1)) {  # priors for all mean treatment effects 
    for (k in (c+1):nt)  {  
	   d[c,k] ~ dnorm(0,.0001)  
	   d[k,c] <- -d[c,k] 
   	   }  
  }   
sd ~ dunif(0,5) # vague prior for between-trial SD. ALTERNATIVES BELOW 
tau <- pow(sd,-2) # between-trial precision = (1/between-trial variance) 
} # *** PROGRAM ENDS

References

1.: Dias, S., Welton, N. J., Sutton, A. J., Caldwell, D. M., Lu, G., Ades, A. E., Evidence Synthesis for Decision Making 4: Inconsistency in Networks of Evidence Based on Randomized Controlled Trials, Medical Decision Making, 33, 641–656, 2013. [PMC free article: PMC3704208] [PubMed: 23804508]
2.: Dias, S., Welton, N. J., Sutton, A. J., Caldwell, D. M., Guobing, L., Ades, A. E., NICE DSU Technical Support Document 4: Inconsistency in networks of evidence based on randomised controlled trials, 2011, last updated April 2014, available from http://scharr.dept.shef.ac.uk/nicedsu/technical-support-documents/evidence-synthesis-tsd-series/ [PubMed: 27466656]
3.: van Valkenhoef, G., Dias, S., Ades, A. E., Welton, N. J., Automated generation of node-splitting models for assessment of inconsistency in network meta-analysis, Research Synthesis Methods, 7, 80–93, 2016 [PMC free article: PMC5057346] [PubMed: 26461181]
4.: Spiegelhalter, D. J., Best, N. G., Carlin, B. P., van der Linde, A. Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society: Series B, 64, 583–616, 2002
5.: Dias, S., Welton, N. J., Caldwell, D. M., Ades, A. E., Checking consistency in mixed treatment comparison meta-analysis, Statistics in Medicine, 29, 932–944, 2010 [PubMed: 20213715]
6.: Dias, S., Ades, A., Sutton, A., Welton, N., Evidence Synthesis for Decision Making 2: A Generalized Linear Modeling Framework for Pairwise and Network Meta-analysis of Randomized Controlled Trials, Medical Decision Making, 33, 607–617, 2013 [PMC free article: PMC3704203] [PubMed: 23104435]
7.: Dias, S., Welton, N. J., Sutton, A. J., Ades, A. E., NICE DSU Technical Support Document 2: A Generalised Linear Modelling Framework for Pairwise and Network Meta-Analysis of Randomised Controlled Trials, 2011, last updated September 2016, available from http://scharr.dept.shef.ac.uk/nicedsu/technical-support-documents/evidence-synthesis-tsd-series/ [PubMed: 27466657]

Footnotes

a: Sterne Jonathan A C, Sutton Alex J, Ioannidis John P A, Terrin Norma, Jones David R, Lau Joseph et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials BMJ 2011; 343 :d4002

Final

Evidence reviews

This evidence review was developed by the Guideline Updates Team

Disclaimer: The recommendations in this guideline represent the view of NICE, arrived at after careful consideration of the evidence available. When exercising their judgement, professionals are expected to take this guideline fully into account, alongside the individual needs, preferences and values of their patients or service users. The recommendations in this guideline are not mandatory and the guideline does not override the responsibility of healthcare professionals to make decisions appropriate to the circumstances of the individual patient, in consultation with the patient and/or their carer or guardian.

Local commissioners and/or providers have a responsibility to enable the guideline to be applied when individual health professionals and their patients or service users wish to use it. They should do so in the context of local and national priorities for funding and developing services, and in light of their duties to have due regard to the need to eliminate unlawful discrimination, to advance equality of opportunity and to reduce health inequalities. Nothing in this guideline should be interpreted in a way that would be inconsistent with compliance with those duties.

NICE guidelines cover health and care in England. Decisions on how they apply in other UK countries are made by ministers in the Welsh Government, Scottish Government, and Northern Ireland Executive. All NICE guidance is subject to regular review and may be updated or withdrawn.

Bookshelf ID: NBK576978PMID: 35099864

PubReader
Print View
Cite this Page
Evidence review for preventing contrast-induced acute kidney injury: Acute kidney injury: prevention, detection and management: Evidence review A. London: National Institute for Health and Care Excellence (NICE); 2019 Dec. (NICE Guideline, No. 148.)
PDF version of this title (9.6M)

In this Page

Preventing contrast-induced acute kidney injury
Appendices

Other titles in this collection

NICE Evidence Reviews Collection

Related NICE guidance and evidence

NICE Guideline 148: Acute kidney injury: prevention, detection and management

Related information

PMC
PubMed Central citations
PubMed
Links to PubMed

Recent Activity

Clear Turn Off Turn On

Evidence review for preventing contrast-induced acute kidney injury
Evidence review for preventing contrast-induced acute kidney injury

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

Evidence review for preventing contrast-induced acute kidney injury

Preventing contrast-induced acute kidney injury

Review question

Introduction

PICO table

Methods and process

Clinical evidence

Included studies

Excluded studies

Quality assessment of clinical studies included in the evidence review

Network meta-analysis

Pairwise meta-analysis

Economic evidence

Included studies

Excluded studies

Summary of studies included in the economic evidence review

Economic model

The committee’s discussion of the evidence

Interpreting the evidence

The outcomes that matter most

The quality of the evidence

NMA analyses and NMA model inconsistency checks

Benefits and harms

Cost effectiveness and resource use

Other factors the committee took into account

Appendices

Appendix A. Review protocols

Review protocol for preventing contrast induced acute kidney injury in at risk adults

Appendix B. Methods

Prioritisation of CI-AKI definitions for pairwise and network meta-analyses

Evidence synthesis and meta-analyses

Evidence of effectiveness of interventions

Quality assessment

Methods for combining intervention evidence (pairwise analysis)

Minimal clinically important differences (MIDs)

GRADE for pairwise meta-analyses of interventional evidence

Publication bias

Methods for combining direct and indirect evidence (network meta-analysis) for interventions

Synthesis

Choice of outcomes for network meta-analysis

Modified GRADE for network meta-analyses

Appendix C. Literature search strategies

Sources searched to identify the clinical evidence

Search strategies

Sources searched to identify economic evaluations

Appendix D. Clinical evidence study selection

Appendix E. Clinical evidence tables

Appendix F. Forest plots of pairwise meta-analysis

Sodium chloride 0.9% vs no (intravenous) hydration

Sodium chloride 0.9% vs oral fluids

Sodium chloride 0.9% vs sodium bicarbonate

Sodium chloride 0.9% + sodium bicarbonate vs sodium chloride 0.9%

Sodium bicarbonate vs no (intravenous) hydration

Oral NAC + sodium chloride 0.45% vs sodium chloride 0.45%

Oral NAC + sodium chloride 0.9% vs sodium chloride 0.9%

Oral NAC + sodium chloride 0.9% vs sodium bicarbonate

Oral NAC + sodium chloride 0.9% vs oral NAC + sodium bicarbonate

Oral NAC + sodium bicarbonate vs sodium chloride 0.9%

Oral NAC + sodium bicarbonate vs sodium bicarbonate

IV NAC + sodium chloride 0.45% vs sodium chloride 0.45%

IV NAC + sodium chloride 0.9% vs sodium chloride 0.9%

IV NAC + sodium chloride 0.9% vs sodium chloride 0.9% + sodium bicarbonate

IV NAC bolus + oral NAC + IV sodium chloride 0.9% vs IV sodium chloride 0.9%

Appendix G. Network meta-analysis results

Model fit statistics

Network diagram

Caterpillar plot

Rank probability histograms

Relative effectiveness

Appendix H. GRADE tables

Pair-wise meta-analysis

Sodium chloride 0.45% vs no (intravenous) hydration

Sodium chloride 0.9% vs no (intravenous) hydration

Subgroup analyses (outcome: CI-AKI): Sodium chloride 0.9% vs no (intravenous) hydration

Other outcomes: Sodium chloride 0.9% vs no (intravenous) hydration

Sodium chloride 0.9% vs oral fluids

Other outcomes: Sodium chloride 0.9% vs oral fluids

Sensitivity analysis excluding studies with a high risk of bias: Sodium chloride 0.9% vs oral fluids

Sodium chloride 0.9% vs sodium chloride 0.45%

Subgroup analyses (outcome: CI-AKI): Sodium chloride 0.9% vs sodium chloride 0.45%